PURPOSE: The SIGNIFICANT (Simple Investigation in Neutropenic Individuals of the Frequency of Infection after Chemotherapy +/- Antibiotic in a Number of Tumours) trial reported a reduction in febrile episodes (FEs) among 1,565 patients with solid cancers and lymphomas receiving cyclical, myelosuppressive chemotherapy (causing grade 4 neutropenia) in a randomized, placebo-controlled, double-blind trial of levofloxacin (P = .01). In response to concerns that increased antibacterial prescribing selects for microbial resistance, we examined our data to explore the rationale for more limited prophylaxis. PATIENTS AND METHODS: The risk of FE was calculated for control patients on first versus nonfirst cycles, with or without first-cycle FE, and within subgroups defined by cancer type, performance status (PS), age, and treatment context (adjuvant v nonadjuvant). Using the randomized trial data, the prophylactic efficacy of levofloxacin was examined for the same subgroups. RESULTS: The per-cycle FE incidence was much lower on nonfirst (3.3%) versus first cycles (8.0%). Prophylaxis was less effective for nonfirst (odds ratio [OR] = 0.78; P = .16) compared with first cycles (OR = 0.42; P <.001). However, FE on cycle 1 predicted a much higher risk of FE and a trend to continued prophylactic efficacy on subsequent cycles. FE rate was greatest for testicular cancer (27.9%), then small-cell lung cancer (17.3%), and lowest for breast cancer (11.5%). Prophylactic efficacy was consistent across age, sex, PS, treatment context, and disease type (except possibly non-Hodgkin's lymphoma). CONCLUSION: Under pressure to limit antibacterial use, these exploratory data support offering prophylactic levofloxacin on cycle 1 only of myelosuppressive cancer chemotherapy and on subsequent cycles after a cycle-1 fever. Prophylactic levofloxacin is effective regardless of age, PS, or tumor type.