The advantage of inoculating blood culture bottles (BCB) with a bulk volume of joint fluid was studied by analyzing results for 155 positive specimens cultured in parallel in BCB and on conventional solid media. The specimens came from both natural and artificial joints of 89 patients treated in 1975-1994. One third of the specimens from patients not on antibiotics and half of the specimens from patients on antibiotics were positive by BCB culture only. Some fastidious or slow-growing organisms were detected exclusively by this method. Additional contaminants were also picked up, but the inconvenience was relatively minor. Alternative procedures for detecting microbes in joint specimens are discussed.