We tested the reliability of a modified version of the Block 98-item food frequency questionnaire (FFQ) and a short 40-item FFQ derived from the Block form to estimate usual calcium intake. Ninety-eight female volunteers completed both forms three times during a 6-month period; forms were initially completed by participants at a group meeting (T1) and, subsequently, 2 months (T2) and 4 months (T3) later, by mail. Both mean total energy intake in kilocalories and mean calcium intake derived from the long FFQ dropped significantly between T1 and T2. The nutrient density variable, calcium per 1,000 kcal, did not vary significantly between administrations. Pearson correlation coefficients were most uniform for pairwise comparisons of calcium per 1,000 kcal at the three times. Mean calcium intake derived from the short FFQ did not vary significantly between administrations, and Pearson correlation coefficients ranged from .70 to .77 for paired comparisons. When using the long FFQ, the nutrient density variable, calcium per 1,000 kcal, was a more reliable measure of usual calcium intake than was calcium in milligrams. The short FFQ provided reliable estimates of usual calcium intake in milligrams and captured more than 75% of the calcium intake measured by the long FFQ.