Much research is never published. The main cause of failure is the poor quality of the data, which does not allow valid conditions to be drawn. This problem can arise because of poor study design. The aims of this presentation are to discuss quality of data and the study designs developed to improve data validity. It is essential to be aware of the many sources of bias, which can fool the inexpert scientist at any stage of the project. These include selection bias, chanelling bias, interview bias, chronology bias, recall bias, citation bias, confounding, and others. These shortcomings can be avoided by proper study design. Different levels of evidence have been defined to categorize study designs. These levels include: ‘expert opinion without critical appraisal’; case-series; case control studies; cohort studies and randomized controlled trials, with systematic reviews in each category being superior to individual studies. Research should be undertaken only by individuals and teams who have a good understanding of study design. Adequate preparation before initiating the study should result in valid data and a good chance of publication in a high-impact journal, if the research findings are indeed new and useful.