Traditionally, measures of psychological symptomology have been employed as the most useful criteria for identifying needs in specific communities, including measures derived from clinical screening inventories. A path analytic model of the relationships among indicators of need was derived from a survey of 1495 residents from four catchment areas of a southwestern state. Measures were obtained of four demographic variables, i.e., age, income, education and divorce status, positive and negative life events, psychological symptoms, life quality, and service use. The most parsimonious causal model was constructed based on the empirical relations among the measures. Life events were found to play an important though complex role in defining need, and life quality as well as symptom measures were found to be useful in predicting service use. Different types of predictors appeared to be important in explaining service use for different catchment areas. The findings suggest that the best indicators of need are, in part, a function of the unique characteristics of the community. (Author/NRB)