BACKGROUND Four previous studies of homeless adults have yielded conflicting results regarding the presence of cognitive impairment.

METHOD A consecutive series of 80 roofless entrants to a hostel for homeless men were sampled and 62 (76%) completed a range of assessments, including measures of mental state, cognitive functions and substance use.

RESULTS Estimated premorbid IQ (mean = 96), current IQ (mean = 84) and cognitive speed were significantly lower than the norm. There was a significant IQ drop in all diagnostic groups. IQ drop, but not current IQ, correlated with duration of rooflessness. Those with schizophrenia or alcohol problems were roofless for longest. Alcohol misuse did not correlate with IQ drop, excepting alcohol withdrawal symptoms in those with schizophrenia.

CONCLUSION The hypothesis that low IQ is a risk factor for rooflessness is supported. However, length of rooflessness was more closely related to IQ drop than to current IQ, suggesting that some third factor may be affecting both rooflessness and intellectual functioning. Roofless men with schizophrenia or alcohol problems may be especially at risk of long-term rooflessness.