Cocoa powder (CP) was fed at levels of 0.0 (control), 1.5, 3.5 and 5.0% for 104 wk to male and female Sprague-Dawley rats derived from the F3b generation of a multigeneration study using the same CP diets. Initial methylxanthine intake was high in all treatment groups, but steadily declined until wk 26. The high dose level provided a mean methylxanthine intake of approximately 57 mg/kg body weight/day for males and 74 mg/kg body weight/day for females from wk 26 to wk 104 of the study. Compared with controls, the historical trend of methylxanthine-associated growth stimulation was evident in rats consuming diets containing 1.5% CP, while body weight was reduced in rats consuming diets containing 3.5 and 5.0% CP. Survival rates were similar in control and CP-fed rats. No evidence of treatment-related clinical disease or ocular effects was noted. An increased incidence of bilateral testicular atrophy and aspermatogenesis was present in males consuming diets containing 5.0% CP. Non-suppurative myocarditis and interstitial fibrosis of the heart were also increased in incidence in both sexes receiving diets containing 5.0% CP. The overall incidences of both pelvic dilatation and renal pelvic microcalculi were increased in most treatment groups. Although there was no difference in the incidence of benign mammary gland fibroadenomas in female rats between the control group and any CP-fed group, a marginally significant (P = 0.04) trend test was apparent. The significance of this finding is doubtful, since the incidence of this lesion in the highest dose group was well within the historical control range for this strain of rats. No evidence of carcinogenicity from dietary CP was found in either sex.