A cohort of 801 studies, submitted to an Australian Institutional Ethics Committee over a 10-year period (Sep 79 - Dec 88), was reviewed and followed to determine the outcome of each study in terms of study results and publication status. The principle aim was to address the question of whether those research studies producing positive or striking results were more likely to be published than studies with negative findings. Of 748 eligible studies, completed questionnaires were obtained from the principal investigators of 520 (70% response): 100 (20%) studies were not commenced, 70 (13%) were abandoned, 63 (12%) were still in progress and 287 (55%) completed. Of the 321 studies with analyses undertaken, 189 (59%) have been published with a median time to publication of 5.5 years. For the 218 studies using quantitative analyses with statistical tests, 99/146 (68%) of 'positive studies' (p<0.05 results) were published compared with 23/52 (44%) of 'negative studies' (p>0.10) and only 4/20 (20%) of 'inconclusive studies' (0.05<p<0.10). Survival analysis of time to publication identified research design (p<0.01), external funding (p<0.06) and statistically significant results (p=0.0004) as independent predictors of a shorter time to publication. The relative risk for publication was 2.34 (95% CI: 1.47-3.73) for statistically significant studies compared with null studies, with a median time to publication of 4.9 vs. 7.7 years. These results were even more striking when restricted to an analysis of 130 relevant clinical trials: relative risk 3.29 (95% CI: 1.84-5.90, p=0.0001). The principle reason for non-publication was failure to submit rather than rejection by journals. These findings confirm those of US and UK studies of publication bias, but also emphasise the delay in time to publication of trials with inconclusive or negative results. This highlights the problem in undertaking a systematic review at any time point without systematically identifying unpublished trials. The study results have stimulated the debate in Australia on the need for the prospective registration of research studies at the time of ethics approval, to ensure all relevant trials are used in systematic reviews.