Background: Randomized control trials and the use of meta-analysis in systematic reviews are the basis of evidence-based practice. The paper reviews their use in the development of evidence-based practice in speech and language therapy. Aims: It is accepted that clinical outcome research should develop in a sequence of phases. A model of this process is described. Examples of outcome research in speech and language therapy are used to illustrate the use of the model and the problems that result when it is not followed. Main Contribution: Existing research has largely ignored the agreed procedures for outcome research. Particular problems have arisen when randomized control trials are used to examine therapy provision for a client group. Clients are often a heterogeneous group and receive different therapies. Consequently, it is unlikely that trials can obtain significant results, nor, if they do, can they provide clinicians with useful information about the choice of treatment. Systematic reviews are equally uninformative. Many of the studies on which they are based have methodological problems and their frequent failure adequately to describe the therapies used mean that reviews cannot evaluate or compare different types of therapy. Conclusions: Researchers in speech and language therapy have given too little attention to the basics of clinical outcome research. This requires that clinical and theoretical insights are used to identify specific therapies for well-defined groups of clients. These therapies must be tested first in efficacy, then in effectiveness studies, and their results should be disseminated to clinicians. Only then is it meaningful to carry out large-scale trials of the effectiveness of therapy provision for a client group or to conduct systematic reviews of existing research.