Our objective was to determine the impact on mortality of the coverage and spectrum of initial empirical antimicrobial treatment among bacteraemic patients in the county of Northern Jutland, Denmark. Patients aged > 15 y from the bacteraemia register who had their first episode of bacteraemia between 1996 and 1998 were retrospectively included. Outcome was defined as survival from the day that the first positive blood culture was taken. Follow-up was 180 d. Survival probability was estimated using a Cox proportional hazard model. Conservative treatment was defined as penicillin G, ampicillin or similar, methicillin or similar, metronidazole, sulfamethoxazole, trimethoprim or a combination of these. We studied 2058 primary episodes of bacteraemia. The distribution of empirical antimicrobial treatments was: conservative/covered, 25.8%; non-conservative/covered, 36.9%; conservative/did not cover, 16.0%; non-conservative/did not cover, 4.8%; and no antimicrobial, 16.5%. After adjusting for risk factors, patients receiving a treatment that covered the microorganism(s) had an increased survival (relative risk [RR] = 0.50 [0.32-0.77]; p = 0.002). Among covered episodes, receiving a conservative treatment significantly increased survival (RR = 0.69 [0.54-0.88]; p = 0.002). This study confirms that appropriate empirical antimicrobial treatment increases the survival of bacteraemic patients. Furthermore, conservative treatment was associated with a longer survival among patients covered by empirical therapy.