Study objectives: We tested the hypotheses that blood culture positivity and contamination rates were not increased by not changing needles between venipuncture and inoculation of blood culture bottles or by taking blood for culture by freshly inserted IV cannulae. Design: A prospective study of blood cultures collected by venipuncture or IV cannulae taken from an emergency department population. Venipuncture samples were randomized into needle change (standard method) or no needle change before inoculation into blood culture bottles. Participants: Nine hundred forty patients requiring blood cultures after assessment in the ED. Interventions: A standard disinfection procedure using 0.5% chlorhexidine in 70% alcohol was used. Blood was collected by venipuncture and inoculated with or without needle change. Blood collected by IV cannula was inoculated with a fresh needle applied to the collection syringe. Measurements and main results: There was no statistically significant difference in contamination rates for blood collected by venipuncture with no needle change (6.4%) compared with needle change (4.2%, P > .30). No significant difference in contamination rates was noted for blood taken by freshly inserted IV cannulae (4.3%) compared with venipuncture with needle change after sampling (4.2%, P > .90). Some problems with randomization resulted in unequal numbers in the needle-change (286) versus no-needle-change (141) subgroups, and this may have introduced bias. A higher rate of pathogen growth was observed in blood taken by IV cannula (11.4%) compared with the standard method (6.3%) ( P < .025). A significantly greater rate of Gram-negative sepsis was noted in the IV cannula group (6.6%) compared with direct venipuncture with needle change (1.1%) and no needle change (4.2%, P < .01). Conclusion: The results of this study do not support the practice of changing needles before inoculating blood samples into blood culture bottles. Collection of blood for culture through freshly inserted IV cannulae is associated with a low contamination rate and is an acceptable alternative to direct venipuncture. Sources of bias in this study suggest that further research is needed to determine the optimal technique for collecting blood cultures.