Operation of domestic smoke detectors (photoelectric and ionization) has been investigated in a typical Melbourne dwelling using smoke from smoldering hardboard and artificial smoke generated at three widely different rates. Detector performance was assessed according to the escape time offered by its operation, this being based on the criterion of visibility needed to effect escape. Under these test conditions detectors located in bedrooms performed adequately only when smoke originated in the same room; they provided poor escape times for smoke originating elsewhere. Also, ionization detectors usually provided adequate escape times only when smoke travel from the source room to the rest of the dwelling was restrained by small door openings, since it then took longer for visibility to be lost. Photoelectric detectors were generally more effective than ionization detectors, as expected for smoldering conditions, and when located in the hallway provided adequate escape times for most conditions of size and location of smoke source.