Calories in, Calories Out (CICO) really stuck for me after experimentation. I managed to verify the model on myself, and on dozens of people over the years. The easy way was controlling people's diets and exercise. That let me see that I could easily have them control their weight. But then, I discovered a simpler, much less involved, and far more scientific way to do this in a blogpost on the topic. The blogpost is entitled "Calories in, calories out" and it shows a simple method you can use to prove to yourself that CICO is real, true, and a great description of reality. To get started, the post says to take your weight and to correctly record reported calories in the food you eat, plus to get down your running mileage for the day. You plug these into a recurrency relation and compute a bunch of predicted weights. When you do this, you eventually get predictions that are right on the mark: Simple enough, no? So, I did this for myself. I used measured my calorie intake, activity, etc., and I started predicting my weight changes with a shocking degree of accuracy. I tried and failed a few times at the outset, but eventually, after I got all my measurements dialed in, it all just worked, and I could even plan my days ahead of time and still get accurate weights predictions. That was proof enough to me that CICO worked. After I found this worked for me, I decided to evangelize. I told people about this, and I applied it to my girlfriend at the time. Turned out, it did not work for her. Odd, I thought, so I decided to monitor her activity and diet more carefully, and I found that what she reported to me was wrong. When I catalogued everything, we had some discrepancies, and when I addressed them, suddenly this worked for her, too! But now I had become acutely aware of the recording issue. If you're unfamiliar, when we actually measure caloric intake using doubly-labeled water, whole-room calorimeters, and other methods like those, there's evidence of systematic misreporting. For example, two recent datasets both showed that fatter people tended to underestimate their caloric intake:* So, I started leading people this by asking them to record their weights on a water fast. That seemed to be a one-shot trick to make the predictions line up really quickly. If you want to get more accurate, the equation you get your predictions from has a burn value per pound of weight. For men, this value tends to be higher than it is for women, in large part because men have a higher proportion of muscle and that tends to burn more calories than fat. You can easily obtain this burn value by measuring your caloric intake and weight and then figuring out the least-squares best-fitting value for the burn rate. The author of the blogposts did this (and provided a means to do it), and it seemed to take him a few weeks to get to a stable value, but eventually he did. With this value personally estimated, you can improve your predictions even further. You can also get to this value faster through leading in with a fast, as that takes out noise from misestimating your caloric intake. Neat, huh? The next question is: Does this hold when you're not losing weight? The answer is, broadly, yes! The author kept recording and eventually got to the point where they went on a trip. They ate at buffets, ate out at restaurants, and were unable to get accurate calorie counts. They also ate enough to regain weight. ...