This paper reports on a project using a series of single subjects to investigate the effectiveness of using electropalatography (EPG) in treating ten children with persisting speech difficulties of no known organic aetiology. The aims of the project were two-fold, firstly to assess whether the subjects selected benefited from this treatment, and secondly to investigate whether it was possible to predict which children would make maximum improvement. A number of factors were identified as possible predictors for successful EPG therapy and subjects were then ranked according to these predictions. Baseline measures of each subject's speech were taken using word lists. Scores reflected the correct number of realizations of consonants produced by each subject. Subjects received the same number of therapy sessions and were then re-tested. Scores before and after therapy were compared and found to be significantly different although the initial predictions as to the magnitude of improvement for each subject were not verified. The selection of appropriate candidates for therapy and the need for objective means of establishing effectiveness are discussed.