The self-report Drug Attitude Inventory (DAI), in 30- and 10-item versions, provides unique information of clinical relevance for monitoring treatment adherence among people diagnosed with schizophrenia. The primary purpose of this paper was to evaluate the 10-item version among patients living in sheltered housing. Data were collected among 68 persons living in sheltered housing, most of them (82%) diagnosed with schizophrenia, 6% with non-organic psychoses, and 12% with other diagnoses. The dichotomic response format of the original DAI-10 was replaced by a 4-point Likert scale, in order to improve the resolution of the scale. Over 90% of the participants produced meaningful scores. A factor analysis suggested a 2-factor orthogonal structure: one highly homogenous factor (5 items) reflected wanted effects of the drug and displayed a bimodal distribution; one factor (3 items) reflected side effects. One item concerned the perceived control over one's drug treatment, which is a key clinical issue. One item was conceptually ambiguous and displayed no correlations with the other items. On the basis of the results we suggest cut-off scores which indicate the need for three kinds of adherence-improving interventions. Summing up, by dropping one item and using a Likert scale response format, the resulting instrument, DAI-9, appears to be an easy-to-use self-report instrument for monitoring drug attitudes and to identify needs for treatment adherence interventions among seriously ill patients.