Research Review: The Dirt On Clean Eating Written By Nutrition Expert Alan Aragon

Everyone knows the difference between dirty and clean foods, so I don’t have to explain the obvious…or do I? My favorite response to questions about how to eat clean is, “Wash your food.” The biggest problem with discussing foods in these terms is that there’s no clear definition of clean or dirty. The difference might seem obvious, but a closer look shows that it’s far from clear-cut. The confusion is compounded when clean eating is preached as the best way to optimal health and body composition.

In this article, I’ll use research and field experience to shed some light on these muddy issues.


The Fickle Nature of Clean

To illustrate the inconsistency of clean through decades, I’ll begin with the 1980’s, widely regarded as the start of the fitness revolution. Through much of the decade, fat (regardless of type) was portrayed by both the academic and lay press as the bad guy. Eating clean in the 80’s was largely characterized by avoiding fat, whether through the plethora of fat-free products, or the vigilant avoidance of all forms of added and naturally occurring fats within foods.

Toward the end of the decade, whole grain products were regarded as the foundation of optimal health.

The 1990’s was a decade that dichotomized unsaturated fats as good, and saturated fats as bad. Red meat, egg yolks, and pretty much all sources of dietary cholesterol were to be avoided. Abundant grain consumption was still encouraged, and even more so if the grain product had a low glycemic index (GI). High insulin elevations were considered harmful to health and body composition.

Therefore, multiple small meals around the clock was recommended not only to control insulin levels, but also to supposedly raise metabolism.

Moderation is Key

Moderation is the key. Gorging on fast foods is most certainly not the way…

Clean in the 2000’s was characterized by the beginnings of amnesty toward saturated fat and cholesterol. They no longer were considered as dirty as previously thought; now hydrogenated vegetable oil was the poison. Omega-3 fatty acids from fish and flaxseed were placed on a heavenly pedestal, receiving the more-is-better stamp. Carbohydrate was now seen as a potentially greater threat to dieters than fat.

Sugar was particularly unclean, as evidenced by the boom of artificially sweetened, low-carb products.

Caveman Clean (Paleo diet)

The present decade has just begun, and eating clean has taken some interesting directions. One is an appeal to imagination about Paleolithic eating habits, which eliminates the consumption of grains, legumes, dairy, added salt, sugar, alcohol, and even certain vegetables. This definition of clean is perhaps the most logically inconsistent one. It emphasizes a prehistoric model, yet many of its proponents take an array of cutting-edge nutritional supplements, and use satellite technology to navigate their drive to the closest parking spot at the gym. Fruits and vegetables have always been a mainstay of clean eating, but pesticide-free produce is now somehow cleaner, pests and all. Another twist in the carbohydrate saga has snowballed as well. Insulin spikes from high-GI carbs were the bane of the 90’s. But now, fructose, a low-GI carbohydrate with minimal effects on insulin response, is now one of the top public enemies.

As you can see, the definition of clean is an elusive target. Are there any common threads among the decades with respect to eating clean? Is there any way to objectively label foods as clean or dirty? Before I get to that, let’s take a look at the concept as it’s been traditionally applied to bodybuilding.

Bodybuilding Clean: The avoidance of Fruits and Dairy

Clean eating in the bodybuilding sense deserves its own discussion. Much of its ‘rules’ are adaptations of dogma from the 80’s and 90’s with a healthy dose of contradiction. Many bodybuilders who consider themselves hardcore will avoid (among other things) dairy and fruit, regardless of training season. Why? Nobody really knows, but I’d speculate that fruit & dairy phobia among bodybuilders originated from the pre-contest leaning-out process, which typically involves the reduction of carbohydrate. Milk and fruit are both carb-dominant foods, and are thus prime candidates for reduction or elimination. But still, my example above is speculative. This dogma could just as easily have come about by someone cutting milk and/or fruit out of the diet and experiencing further fat loss from the re-creation of an energy deficit, and declaring those foods barriers to fat loss. Nevertheless, in some pre-contest cases, carbohydrate restriction to extreme degrees is called for, and this nullifies the possibility of including milk & fruit (or any carb source, for that matter), at least cyclically. So, milk and fruit got blamed as bad for all occasions, when their omission only potentially applies to certain aggressively carb-restricted dieting phases.

Bodybuilders often pride themselves on having nutrient-rich diets, yet many of them opt for a significant portion of their day’s carbohydrate allotment as dextrose (or some other empty-calorie carb source) instead of fruit.


Attempts at Objectively Defining Clean

Scientific investigations of the nutritional status of bodybuilders have shown some interesting results, and here are some of the highlights. Kleiner and colleagues examined the pre-contest dietary habits of male & female junior national & national-level competitors,15-40% of whom admitted to using various drugs [1]. Despite consuming adequate total calories, women were “remarkably deficient” in calcium intake, which is not surprising given the widespread milk-phobia among bodybuilders. In subsequent work led by Kleiner on female & male competitors at the first drug-tested USA Championship, men consumed only 46% of the RDA for vitamin D. Women consumed 0% of the RDA for vitamin D, and 52% of the RDA for calcium [2]. Zinc, copper, and chromium were also underconsumed by the women. Despite dietary magnesium intakes above the RDA, serum magnesium levels in females were low. Serum zinc levels were high in men and women. It’s notable that not all research on bodybuilders has found nutrient deficiencies. Intakes in significant excess of the RDA in both offseason and pre-contest conditions have also been seen [3,4].

Still, the potential for nutrient deficiencies in this population is strong due to the elimination of food groups combined with a high training volume and lowered caloric intake overall.

Whole vs Processed: Is your Whey Dirty?

The two most commonly cited characteristics of foods considered clean are a lack of processing and a high nutrient density. Let’s look at processing first. Foods in their whole, naturally occurring state are often deemed clean. In contrast, foods that are altered or removed from their original state are stripped of the clean stamp. Is this demerit warranted? As we’ll see, this is not a reliable method of judgment for all foods. By this definition, most supplements are dirty, since they often undergo extensive processing and are far-removed from their original source. To use a common example, whey is doubly processed in the sense that it’s not only a powdered form of milk protein, but it’s a separated fraction of milk protein. Yet, when combining the results of standard ranking methods (biological value, protein efficiency ratio, net protein utilization, and protein digestibility corrected amino acid score), whey has a higher total than all other proteins tested, including beef, egg, milk, and soy [5].

Furthermore, research has shown not only its benefits for training applications [6], but whey has a surprisingly wide range of potential for clinical applications as well [7-10]. Therefore, despite whey being a refined/processed food, it has multiple benefits and minimal downsides.

Nutrient Density: A qualifier for “Clean” foods?

The next commonly proposed qualifier for a food to be considered clean is its nutrient density. A little-known fact is that there is no scientific consensus on what nutrient density actually means. To quote Miller and colleagues [11],“There is currently no science-based definition for either nutrient density or nutrient-dense foods.

Without a definition that has been developed using an objective, scientific approach, the concept of what is a “nutritious” food is subjective and, therefore, inconsistent.”

The existence of multiple methods of measuring diet quality illustrates the point expressed in the quote above. Nutrient profiling systems include the Healthy Eating Index (HEI), Diet Quality Index, and Alternative HEI. The most recent profiling method is the Nutrient Rich Foods Index (NRFI). The NRFI attempts to consolidate principles from previous methods to establish a more comprehensive definition of nutrient density.

It judges individual foods based on the presence of selected important nutrients and absence of problematic ones [12]. Still, the NRFI has its bugs and biases, particularly against saturated fat (& fat in general).

GSW : A Simplistic approach

A simplistic learning tool called the “Go, Slow, and Whoa” (GSW) food classification system was designed to help children and families make better food choices [13]. GSW was recently compared with the more sophisticated NRFI, and despite some differences, both methods closely corresponded with each other in terms of distinguishing energy-dense and nutrient-rich foods [14]. Although the two methods aligned fairly well, they also share similar out-dated ideologies. For example, sports drinks have a “Slow” designation, and whole milk is nailed as a “Whoa” food – brilliant, huh? Tuna canned in water is in the most favorable “Go” column, while fatty fish like salmon is not even listed.

A final example is the listing of egg whites in the “Go” column, and whole eggs in the “Slow” column. Unsurprisingly, the government-issued guidelines are still stuck in the fat-phobic era.

Perils of Judging the Parts & Not the Whole

3In the process of classifying foods based on nutrient density, the context of the foods within the diet as a whole is often lost. Attempts at defining nutrient density of foods on an individual basis, for the most part, have failed. Much of the classifications are out-dated at best, and counterproductive at worst. It would seem to be a simple matter of labeling foods with a high ratio of micronutrients to calories as nutrient-dense, and foods with a high ratio of calories to micronutrients as energy-dense. However, this simply is not the case. An energy-dense food can still contain more essential macronutrition and/or bioavailable micronutrition than a nutrient-dense, energy-sparse food. Another thing that tends to get ignored is that athletes with high endurance demands or high overall training volume would compromise their performance if energy density was neglected.

Ultimately, it’s impossible to judge a food in isolation from the rest of the diet. Furthermore, it’s impossible to judge a diet without considering the training protocol, goals, preferences, and tolerances of the individual.

Dirty Fat Loss (Carbohydrates are not the enemy)

Clean diets are commonly touted to produce more favorable body composition changes than unclean diets. In fact, some even claim that dirty dieting will not allow fat loss to occur. For weight or fat loss, concerns of a dirty diet used to be centered on fat intake. That’s no longer the case; carbohydrate has been receiving the brunt of the contempt lately. In light of the current sugar-phobic climate with an emphasis on fructose, the following studies deserve special attention.

First up, Surwit and colleagues compared the 6-week effects of 2 hypocaloric diets – one with 43% of the total calories as sucrose (table sugar), and one with 4% of the total calories as sucrose [15].

No significant differences were seen in the loss of bodyweight or bodyfat between the high and low-sucrose groups. Strengthening these results was the use of dual X-ray absorptiometry (DXA) to measure body composition. Furthermore, no differences in blood lipids or metabolism were seen between the groups. It looks like a more sugary intake still cannot override a calorie deficit. Next up is a recent study by Madero and colleagues, comparing the 6-week effects of a low-fructose diet (less than 20 g/day) or a moderate-fructose diet (50-70 g/day) mostly from whole fruit [16]. The moderate-fructose group lost significantly more weight than the low-fructose group (4.19 kg versus 2.83 kg, respectively). Notably, the moderate-fructose group lost slightly more fat, but not to a statistically significant degree.

Unfortunately, body composition was measured with bioelectrical impedance analysis (BIA) instead of something more reliable like DXA. Nevertheless, bodybuilders afraid of fruit would have to admit that the dirtier diet prevailed in this case.

Trans fatty acids (TFA) have earned a lot of bad press for their adverse effects on biomarkers of cardiovascular health [17,18]. However, some research indicates that not all TFA are harmful. A distinction should be made between industrially produced TFA via hydrogenation of vegetable oils, and naturally occurring TFA in dairy and meat [19]. Vaccenic acid, the main form of TFA in ruminant fats, might actually lower the risk for coronary heart disease [20]. Currently, there’s no controlled human research specifically comparing the effects of TFA with other types of fats on body composition.

In any case, the fitness-conscious population has nothing to worry about unless they start indiscriminately gorging on fast food, cooking with vegetable shortening, and pounding loads of processed/packaged pastries and desserts.

All-or-Nothing Dieting & Eating Disorder Risk

In 1997, a general physician named Steven Bratman coined the term orthorexia nervosa [21], which he defines as, “an unhealthy obsession with eating healthy food.” It reminds me of the counterproductive dietary perfectionism I’ve seen among many athletes, trainers, and coaches. One of the fundamental pitfalls of dichotomizing foods as good or bad, or clean or dirty, is that it can form a destructive relationship with food. This isn’t just an empty claim; it’s been seen in research. Smith and colleagues found that flexible dieting was associated with the absence of overeating, lower bodyweight, and the absence of depression and anxiety [22]. They also found that a strict all-or-nothing approach to dieting was associated with overeating and increased bodyweight. Similarly, Stewart and colleagues found that rigid dieting was associated with symptoms of an eating disorder, mood disturbances, and anxiety [23]. Flexible dieting was not highly correlated with these qualities.

Although these are observational study designs with self-reported data, anyone who spends enough time among fitness buffs knows that these findings are not off the mark.

Applying Moderation: The 10-20% Guideline

For those hoping that I’ll tell you to have fun eating whatever you want, you’re in luck. But, like everything in life, you’ll have to moderate your indulgence, and the 10-20% guideline is the best way I’ve found to do this. There currently is no compelling evidence suggesting that a diet whose calories are 80-90% from whole & minimally processed foods is not prudent enough for maximizing health, longevity, body composition, or training performance. As a matter of fact, research I just discussed points to the possibility that it’s more psychologically sound to allow a certain amount of flexibility for indulgences rather than none at all. And just to reiterate, processed does not always mean devoid of nutritional value.

Whey and whey/casein blends are prime examples of nutritional powerhouses that happen to be removed from their original food matrix.

Use the 10-20% discretionary intake rule and enjoy life a bit.

The 10-20% guideline isn’t only something I’ve used successfully with clients; it’s also within the bounds of research. Aside from field observations, there are three lines of evidence that happen to concur with this guideline. I’ll start with the most liberal one and work my way down. The current Dietary Reference Intakes report by Food & Nutrition Board of the Institute of Medicine lists the upper limit of added sugars as 25% of total calories [24]. Similarly, an exhaustive literature review by Gibson and colleagues found that 20% of total calories from added sugars is roughly the maximum amount that won’t adversely dilute the diet’s concentration of essential micronutrition [25]. Keep in mind that both of these figures are in reference to refined, extrinsic sugars, not naturally occurring sugars within whole foods like fruit or milk. Finally, the USDA has attempted to teach moderation with their concept of the discretionary calorie allotment, defined as follows [26]:

“…the difference between total energy requirements and the energy consumed to meet recommended nutrient intakes.”


Basically, discretionary calories comprise the margin of leftover calories that can be used flexibly once essential nutrient needs are met.

Coincidentally, the USDA’s discretionary calorie allotment averages at approximately 10-20% of total calories [27]. Take note that discretionary calories are not just confined to added sugars. Any food or beverage is fair game. The USDA’s system is still far from perfect, since it includes naturally-occurring fats in certain foods as part of the discretionary calorie allotment.

This is an obvious holdover from the fat-phobic era that the USDA clings to, despite substantial evidence to the contrary [28].

It’s important to keep in mind that protein and fat intake should not be compromised for the sake of fitting discretionary foods into the diet. In other words, make sure discretionary intake doesn’t consistently displace essential micro- & macronutrient needs, and this includes minimum daily protein and fat targets, which vary individually. This may be tough to accept, but alcohol is not an essential nutrient.

Its risks can swiftly trump its benefits if it’s consumed in excess, so it falls into the discretionary category.

10% Versus 20%

Another legitimate question is why I’ve listed the discretionary range as 10-20% rather than just listing it as a maximum of 20%. This is because energy balance matters. In bulking scenarios, maintaining a 20% limit could potentially pose health risks that are already elevated by the process of weight gain, which in some cases involves a certain amount of fat gain. Conversely, weight loss tends to be an inherently cardioprotective process, independent of diet composition [29]. So, the 20% limit is more appropriate for those either losing or maintaining weight. Those who are gaining weight but want to play it safe should hover towards the lower & middle of the range (10-15%). Another factor that can influence the upper safe threshold is physical activity level. I’ll quote Johnson & Murray in a recent review [30]:

“Obesity and metabolic syndrome are rare among athletes, even though dietary fructose intake is often high, underscoring the robust protective role of regular exercise.”

4In the above quote, you can substitute any controversial food or nutrient in place of the word fructose, and the same principle would apply. A greater range of dietary flexibility is one of the luxuries of regular training. Sedentary individuals do not have the same level of safeguarding from the potentially adverse effects of a higher proportion of indulgence foods. And just in case it wasn’t made clear enough, 10-20% indicates the maximum, not minimum discretionary allotment. If someone strives to consume 0% of calories from any food that’s been processed or refined from its original state, then that’s perfectly fine – as long as this is the person’s genuine preference, and not a painful battle of will. I’d also like to make it clear that there is still plenty of grey area in the study of dietary effects on health.

As such, the nature and extent of the miscellaneous or rule-free food allotment is a delicate judgment call. In this case, it’s wise to keep scientific research at the head of the judging panel, but don’t ignore personal experience & individual feedback.

Final Note: Linear Versus Nonlinear Distribution

A legitimate question is, what’s the best way to distribute discretionary calories? Should they be confined to a daily limit, or can it be a weekly limit? The best answer is to let personal preference decide. If we use a 2000 kcal diet as an example, a flat/linear approach would mean that 200-400 kcal per day can come from whatever you want, while meeting essential needs otherwise in the diet. Weekly, this translates to 1400-2800 kcal, depending on the factors I previously discussed. One nonlinear option would be to break the weekly allotment in half, where 2 days per week you indulge in 700-1400 kcal of whatever you want, keeping the remaining 5 days relatively Spartan. Again, there is no universally superior method of distributing the discretionary allotment. The same principle applies to the choice of foods to fulfill it.

Honoring personal preference is one of the most powerful yet underrated tactics for achieving optimal health and body composition. And that’s the nitty-gritty as I see it.

About the Author

Alan Aragon has over 20 years of success in the fitness field. He earned his Bachelor and Master of Science in Nutrition with top honors. He is a continuing education provider for the Commission on Dietetic Registration, National Academy of Sports Medicine, and National Strength & Conditioning Association. Alan has lectured to clinicians at the FDA and the annual conference of the Los Angeles Dietetic Association. He maintains a private practice designing programs for recreational, Olympic, and professional athletes,including the Los Angeles Lakers, Los Angeles Kings, and Anaheim Mighty Ducks.

Article submitted by:
Author’s website:


1. Kleiner SM, et al. Metabolic profiles, diet, and health practices of championship male and female bodybuilders. J Am Diet Assoc. 1990 Jul;90(7):962-7.
2. Kleiner SM, et al. Nutritional status of nationally ranked elite bodybuilders. Int J Sport Nutr. 1994 Mar;4(1):54-69.
3. Keith RE, et al. Nutritional status and lipid profiles of trained steroid-using bodybuilders. Int J Sport Nutr. 1996 Sep;6(3):247-54.Hoffman JR, Falvo MJ. Protein-which is best? J Sport Sci Med 2004; 3: 118-30.
4. Bamman MM, et al. Changes in body composition, diet, and strength of bodybuilders during the 12 weeks prior to competition. J Sports Med Phys Fitness. 1993 Dec;33(4):383-91.
5. Hoffman JR, Falvo MJ. Protein-which is best? J Sport Sci Med 2004; 3: 118-30.
6. Hulmi JJ, et al. Effect of protein/essential amino acids and resistance training on skeletal muscle hypertrophy: A case for whey protein. Nutr Metab (Lond). 2010 Jun 17;7:51.
7. Xu R. Effect of whey protein on the proliferation and differentiation of osteoblasts. J Dairy Sci. 2009 Jul;92(7):3014-8.
8. Krissansen GW. Emerging health properties of whey proteins and their clinical implications. J Am Coll Nutr. 2007 Dec;26(6):713S-23S.
9. Parodi PW. A role for milk proteins and their peptides in cancer prevention. Curr Pharm Des. 2007;13(8):813-28.
10. Marshall K. Therapeutic applications of whey protein. Altern Med Rev. 2004 Jun;9(2):136-56.
11. Miller GD, et al. It is time for a positive approach to dietary guidance using nutrient density as a basic principle. J Nutr. 2009 Jun;139(6):1198-202.
12. Fulgoni VL 3rd, et al. Development and validation of the nutrient-rich foods index: a tool to measure nutritional quality of foods. J Nutr. 2009 Aug;139(8):1549-54.
13. US Department of Health and Human Services, National Heart Lung and Blood Institute. We can! Go, Slow and Whoa foods.
14. Drewnowski A, Fulgoni V 3rd. Comparing the nutrient rich foods index with “Go,” “Slow,” and “Whoa,” foods. J Am Diet Assoc. 2011 Feb;111(2):280-4.
15. Surwit RS, et al. Metabolic and behavioral effects of a high-sucrose diet during weight loss. Am J Clin Nutr. 1997 Apr;65(4):908-15.
16. Madero M, et al. The effect of two energy-restricted diets, a low-fructose diet versus a moderate natural fructose diet, on weight loss and metabolic syndrome parameters: a randomized controlled trial. Metabolism. 2011 May 27. [Epub ahead of print]
17. Mozaffarian D, Clarke R. Quantitative effects on cardiovascular risk factors and coronary heart disease risk of replacing partially hydrogenated vegetable oils with other fats and oils. Eur J Clin Nutr. 2009 May;63 Suppl 2:S22-33.
18. Wallace SK, Mozaffarian D. Trans-fatty acids and nonlipid risk factors. Curr Atheroscler Rep. 2009 Nov;11(6):423-33.
19. Chardingny JM, et al. Do trans fatty acids from industrially produced sources and from natural sources have the same effect on cardiovascular disease risk factors in healthy subjects? Results of the trans Fatty Acids Collaboration (TRANSFACT) study. Am J Clin Nutr. 2008 Mar;87(3):558-66.
20. Field CJ, et al. Human health benefits of vaccenic acid. Appl Physiol Nutr Metab. 2009 Oct;34(5):979-91.
21. Bratman S. What is orthorexia? Accessed August 2011.
22. Smith CF, et al. Flexible vs. Rigid dieting strategies: relationship with adverse behavioral outcomes. Appetite. 1999 Jun;32(3):295-305.
23. Stewart TM, et al. Rigid vs. flexible dieting: association with eating disorder symptoms in nonobese women. Appetite. 2002 Feb;38(1):39-44.
24. Food and Nutrition Board, Institute of Medicine. Dietary Reference Intakes for Energy, Carbohydrate, Fiber, Fat, Fatty Acids, Cholesterol, Protein, and Amino Acids. 2005.
25. Gibson SA. Dietary sugars intake and micronutrient adequacy: a systematic review of the evidence. Nutr Res Rev. 2007 Dec;20(2):121-31.
26. DGAC Advisory Committee, USDA. Part D, Section 3: Discretionary Calories. The Report of the Dietary Guidelines Advisory Committee on Dietary Guidelines for Americans, 2005.
27. Center for Nutrition Policy & Promotion. My Pyramid: Food intake patterns, 2005.
28. Hession M, et al. Systematic review of randomized controlled trials of low-carbohydrate vs. low-fat/low-calorie diets in the management of obesity and its comorbidities. Obes Rev. 2009 Jan;10(1):36-50.
29. Leenen R, et al. Relative effects of weight loss and dietary fat modification on serum lipid levels in the dietary treatment of obesity. J Lipid Res. 1993 Dec;34(12):2183-91.
30. Johnson RJ, Murray R. Fructose, exercise, and health. Curr Sports Med Rep. 2010 Jul-Aug;9(4):253-8.