Non-steroidal anti-inflammatory drugs (NSAIDs), especially acetyl salicylic acid (ASA), associate with reduced breast cancer (BCa) mortality. COX-2 overexpression is tied to worse BCa prognosis, but COX-2 inhibitors (coxibs) have not been shown to have survival advantage. However, the evidence is conflicting and limited. We examined the association between BCa mortality and NSAID use in a Finnish population-based cohort. The study cohort, 73,170 females diagnosed with BCa during 1995-2013 was identified from The Finnish Cancer Registry. NSAID purchases during 1995-2015, the time and causes of death, mammographic screening participation and tumor hormone receptor status were obtained from national registries. Separate exposure variables were created for purchases of all NSAIDs combined, ASA and coxibs . Multivariable-adjusted Cox proportional hazard regression was used to compare BCa-specific and overall survival by the use of NSAIDs, ASA and coxibs compared to non-users. Pre-diagnostic use of any NSAIDs (HR 0.78, 95% CI 0.75-0.81) and coxibs (HR 0.76, 95% CI 0.71-0.81), but not ASA were associated with lowered BCa mortality. Conversely post-diagnostic use of any NSAIDs was associated with increased BCa mortality (HR 1.27, 95% CI 1.22-1.33), while ASA use (HR 0.84, 95% CI 0.73-0.97) showed dose-dependent risk reduction. Post-diagnostic use of ASA associated with reduced BCa-specific mortality, distinguishing ASA from other NSAIDs. Clinical trials should be performed to confirm this association and to determine ideal post-diagnostic ASA dose, frequency, and duration for improving BCa-specific survival. The significance of pre-diagnostic NSAID use as a prognostic factor in BCa warrants further investigation.