This study reviews the extent of evaluation of health outcomes in three general medicaljournals over the past decade by examining papers published in the original researchsection of the New England Journal of Medicine (NEJM), The Lancet, and theMedical Journal of Australia (MJA) in 1982 and 1992. Evaluations were identifiedand classified according to the type of comparison group and the type of outcomemeasures employed. They were divided into three categories: those employing acomparison group; those employing a before-and-after study design (or owncomparison group); and those with no comparison group. The categories of outcomemeasures were mortality, clinical or intermediate measures of health state, and finaloutcome measures (quality of life). Results show that the proportion of papersevaluating a health services intervention remained stable over the period. However,the MJA published considerably fewer evaluations than the other journals. In theNEJM and The Lancet, 75- per cent of evaluations incorporated comparison groups,in the MJA, less than 40- per cent. Overall, the proportion of papers reporting finaloutcome measures increased significantly between 1982 and 1992 (p = 0.04) butthe change in each journal individually did not reach statistical significance. Thisstudy indicates that the reporting of health outcomes evaluations has remainedconstant but there has been some change in the use of comparison groups and finaloutcome measures over time.