• A series of three experiments addressed major problems concerning the use of serum levels as predictors of clinical response to thiothixene (Navane) hydrochloride in schizophrenia: correlation of initial test doses with clinical response; comparison of fluorescence spectrophotometry with gas chromatography in relation to clinical response; and comparison of serum levels with RBC levels in relation to clinical response. All assays correlated (near r=.5) with Brief Psychiatric Rating Scale improvement during hospitalization, except RBC levels seemed to have superior correlations (.64) in patients with lower serum levels. These correlations are similar to those obtained with steady state levels. The different methods of determining thiothixene concentrations were highly intercorrelated as well. Thus, single-dose serum levels give important clinical correlations regardless of which assay is used for thiothixene determination.