A survey of instructional, administrative, and policy-level personnel who participated in needs assessment studies in 12 Florida school districts revealed several problems in the use of the discrepancy model in needs assessment. Nine of the district studies focused on instructional needs, such as basic skills or early childhood education; two studies addressed inservice training needs; and one district study constituted a comprehensive needs assessment for long-range planning of instructional, managerial, and inservice programs. Within each of these 12 districts, questionnaires were sent to 24 school board members, 47 district-level administrative staff members, 43 school principals, and 36 teachers who had participated in a needs assessment study in the previous year. The response rate was 70% for board members, 87% for district-level administrators, 72% for principals, and 54% for teachers. The problems uncovered relate to the use of size of discrepancy as a basis for establishing needs priorities and to the setting of priorities without adequate consideration of potential solutions. Alternative criteria for setting needs priorities are presented. Needs assessments based on the importance of goals and size of discrepancy were effective in producing a list of needs ranked by priority; they were less effective in bringing about change. The limited use of needs assessment results was due, in part, to low confidence in the reliability of priorities based on size of discrepancy. Many practitioners preferred change-related criteria and suggested that integrating needs assessment processes with existing planning processes would improve the validity of decisions made about both needs and solutions. One data table is included. (TJH)