The accuracy of previous equations for predicting resting metabolic rate (RMR) in healthy older men is questionable because they are based on limited sample sizes and the absence of cross-validation procedures. The purposes of this study were to (1) examine biological predictors of RMR in healthy older men; (2) develop a practical equation to predict RMR from easily measured variables and examine its accuracy using cross-validation procedures; and (3) test the validity of existing equations in the literature to predict RMR in older men by comparison with measured RMR values. RMR, body composition, anthropometric measurements, leisure time activity (LTA), maximal aerobic power (VO2max), energy intake, and plasma thyroid hormone levels were determined in 89 healthy older men aged 50 to 78 years. Stepwise regression analysis showed that RMR was best predicted by fat-free weight ([FFW] R2 = 85%), free 3,5,3'-triiodothyronine (T3) level (R2 = 1%), and VO2max (R2 = 1%); these variables predicted RMR with a residual error of +/- 30 kcal/d. A practical equation was developed in a randomly selected subsample (N = 61) using easily measured variables as potential predictors, and was successfully cross-validated in a random subsample of older men (N = 28). The pooled equation to predict RMR is as follows: RMR (in kilocalories per day) = 9.7 (weight in kilograms) - 6.1 (chest skinfold thickness in millimeters) - 1.8 (age in years) + 0.1 (leisure time activity [LTA] in kilocalories per day) + 1,060. These variables accounted for 76% (R2) of the variation, and predicted RMR with a residual error of +/- 42 kcal/d.(ABSTRACT TRUNCATED AT 250 WORDS)