To allow appropriate allocation of prevention and care funding, HIV/AIDS surveillance data must include risk factor information, currently available for less than 70% of cases reported in the United States. The authors evaluated an intervention consisting of provider training and materials to improve risk factor reporting. Facilities were matched prior to randomization to intervention or control, and generalized linear mixed models were used to test for an intervention effect. Twenty-one percent of cases from intervention facilities and 33.4% from control facilities (p = 0.09) were reported without any risk factor information. The pre-post difference (20.7% for intervention and 36.0% for control) was not significant among HIV cases (p = 0.11) nor among AIDS cases (p = 0.12; 21.3% for intervention and 31.1% for control). The methods the authors' evaluated may need to be combined with other approaches and/or alternative classification schemes to significantly reduce the percentage of cases reported to surveillance without risk factor information. (Contains 4 tables.)