This study assesses how social desirability affects responses in clinical self-report inventories. Six hundred items gathered from four normal personality questionnaires were adapted to devise a pre-experimental personality questionnaire (pre-EMHQ). Results obtained from administering Ko's Mental Health Questionnaire (KMHQ) and the pre-EMHQ to separate samples were the proportion of individuals answering "true" to each item (i.e., P(t)) and the social desirability scale value (i.e., SDSV) of each item. The Experimental Mental Health Questionnaire (EMHQ) was established from the pre-EMHQ by closely matching the P(t)s and the SDSVs of the two questionnaires. Administering the KMHQ and the EMHQ concurrently to another sample provided results for factor analysis and other statistical analyses. The SDSVs and the P(t)s for each of the KMHQ items certainly displayed a linearly increasing relation. The two sets of corresponding subscales also correlated significantly. By applying the polynomial regression analysis, the tendency to score might be expressed as a quadratic function of SDSVs. Two iterative principal-factor analyses of the two sets of subscales each resulted in two factors, and Factor 1 is similar in both the KMHQ and the EMHQ. In brief, social desirability plays a critical role in affecting responses in a clinical self-report inventory. The factors involved and suggestions proposed will be of value for further research.