An 84-item, semiquantitative dietary history questionnaire was developed for epidemiologic studies of diet and diseases in Chinese Americans. The questionnaire is a 30-minute, face-to-face administered instrument designed especially for assessing mixed dishes; it consists of a food list and requires information on frequency of consumption and portion estimation. A nutrient database was compiled from many sources, but mostly from US Department of Agriculture publications. The relative validity of this food frequency questionnaire was tested among 74 Chinese women between the ages of 30 and 60 years living in the San Francisco Bay area against a 1-day recall deemed a typical day's diet. Correlation coefficients of nutrients from both methods ranged from 0.2 for total fat to 0.7 for calcium and were all statistically significant at less than the 0.001 level. Agreement between these two methods was assessed by the percentage of misclassification of quartile distributions. Around 10% were grossly misclassified and 50% were correct estimates of intake in the same highest quartile, while 73% were correctly placed in the two highest quartiles. This food frequency questionnaire can reasonably estimate the usual dietary intakes among Chinese Americans for epidemiologic studies, but further validation would be needed.