Eighty-five patients, age 48 to 77 y with postmenopausal crush fracture osteoporosis, were investigated using a 7-d combined calcium balance and 47Ca tracer-kinetic turnover method taking the dermal calcium loss into account. Individual dietary records were obtained based on a 4-d registration at home before hospital admission and on questioning by a dietitian. The following dietary constituents were estimated: energy content, protein, methionine, cysteine, calcium, phosphate, magnesium, coffee, fiber and vitamin C. All patients were served individual diets based on the dietary records during study. Dietary calcium was measured in duplicates of all the meals served. All urine and feces were collected and analyzed for calcium content. The 47Ca kinetic data were analyzed according to a modification of the expanding calcium pool model. The overall calcium balance correlated significantly to energy content (r = 0.31, P less than 0.005), protein (r = 0.22, P less than 0.05), calcium (r = 0.28, P less than 0.01), phosphate (r = 0.27, P less than 0.02) and coffee (r = -0.21, P less than 0.05). However, a multiple backward linear regression analysis disclosed that only calcium (rp = 0.38, P less than 0.0005) and coffee intake (rp = -0.25, P less than 0.05) significantly influenced calcium balance. The equation was: calcium balance (mmol/d) = 0.14 x (dietary calcium, mmol/d) - 0.0016 x (coffee intake, mL/d) - 3.62. A coffee intake in excess of 1000 mL could induce an extra calcium loss of 1.6 mmol calcium/d, whereas intakes of 1-2 cups of coffee per day would have little impact on calcium balance.