Objectives Many residential fire deaths occur in homes with no or non-functioning smoke alarms (SAs). Self-reported SA coverage is high, but studies have found varying validity for self-report measures. The authors aim to: (1) determine over-reporting of coverage, (2) describe socio-demographic correlates of over-reporting and (3) report reasons for over-reporting.

Methods The authors surveyed 603 households in a large, urban area about fire safety behaviours and then tested all SAs in the home. 23 participants who over-reported their SA coverage were telephoned and asked about why they had misreported.

Results Full coverage was reported in 70% of households but observed in only 41%, with a low positive predictive value (54.2%) for the self-report measure. Most over-reporters assumed alarms were working because they were mounted or did not think a working alarm in a basement or attic was needed to be fully protected.

Conclusions If alarms cannot be tested, researchers or those counselling residents on fire safety should carefully probe self-reported coverage. Our findings support efforts to equip more homes with hard-wired or 10 year lithium battery alarms to reduce the need for user maintenance.