We characterized the accuracy, sensitivity, and reliability of computed tomographic (CT) estimates of intra-abdominal (IA) and subcutaneous (S) adipose tissue (AT) in rats and mice using the Aloka rodent CT. Here, we present the first comparisons of CT estimates of the weights of AT samples ex vivo to balance weights of the same samples, of CT estimates of AT weights in vivo to the weights of resected whole-body AT, and of CT estimates of the weights of pieces of AT inserted IA or S in vivo to the weights of the same pieces ex vivo. CT underestimated AT weight ex vivo by approximately 10%, and correction of the automated categorization of IAAT and SAT by Aloka software was required. After these adjustments, correlations (r) of CT estimates and balance weights of resected AT were > or =0.99 in rats and > or =0.92 in mice. CT was impressively sensitive: the 95% probability range of CT estimates of 10,000 mg AT inserts into rats was +/-780 mg and of 500 mg inserts into mice, +/-20 mg. Scans limited to the abdominal region correlated well (r > 0.90) with whole-body scan measures of IAAT and SAT in rats and with IAAT, but not SAT (r < 0.80), in mice. Sums of IAAT and SAT correlated well with body weight in rats (r > 0.90), but not in mice (r < 0.80). Coefficients of variance (CVs) of duplicate scans were <5%. We conclude that CT is a valid tool for studies of AT weight in rats and mice, especially when rapid throughput or longitudinal measures are desired.