The effect of delay in processing on results of lysis-centrifugation (LC; Isolator) blood cultures was assessed in 4,577 paired blood specimens. Blood specimens were obtained at all hours from 384 febrile marrow transplant patients with indwelling venous catheters and were processed by the LC technique and by a conventional two-bottle method. Most patients (84%) were receiving broad-spectrum antibiotics at the time of blood culture. Specimens were delivered to the laboratory, where Isolator tubes were held at 35 degrees C and processed in batches between 0700 and 1730 h daily. This procedure resulted in a delay beyond the manufacturer-suggested processing time of less than 8 h for 1,853 (42%) of the LC cultures. There was no overall difference in the recovery of organisms present in LC cultures processed after being held for 8 to 24 h compared with the conventional two-bottle method. LC methodology had shorter time to detection than the conventional method for detection of Candida spp. and Pseudomonas spp. (P less than 0.05). However, time to detection for Streptococcus spp. and members of the family Enterobacteriaceae, responsible for 16.3% of total isolates, was prolonged significantly by delay in processing when compared with the conventional two-bottle method (P less than 0.01). Results of this study support the recommendation of the manufacturer for processing of Isolator tubes within 8 h or less. Although one can safely delay processing beyond 8 h in terms of total recovery of organisms, such delays were associated with longer time to detection for certain important potentially pathogenic organisms which accounted for a sizeable proportion of blood culture isolates from marrow transplant patients.