In adult chickens, feed intake can be predicted by multiple linear regression from body weight, change in body weight during the recording period (for males and females), and egg mass (for females). Residual feed consumption (RFC) is estimated by the deviation of observed from predicted values for feed intake. A divergent selection experiment has been conducted since 1975 in a Rhode Island Red population. Each sex was selected on the basis of its own RFC, Line R+ for high values of RFC and Line R- for low values. In addition to the traits measured to obtain RFC, egg production traits and body measurements (BM) have been recorded, including wattle length, shank length, and rectal temperature. After pooling data from the two lines and the base population, the data set included 1,064 males, each with 7 variables, and 3,780 females, each with 11 variables. Genetic parameters were estimated by a multivariate derivative-free-restricted maximum likelihood procedure, which yields estimates free of bias due to selection and inbreeding. The RFC appeared to be moderately heritable in males (h2 = .33) and in females (h2 = .27) and poorly correlated between sexes, with a genetic correlation of .19 between RFC in males (RFCm) and in females (RFCf). The RFCm and RFCf were weakly correlated with egg production traits. Significant, positive correlations were found between RFCm and the BM traits that were related to heat dissipation. Correlations between feed intake, body weight, and BM traits recorded in males or in females were always < 1, except for shank length.(ABSTRACT TRUNCATED AT 250 WORDS)