Objectives  To determine if a period in a research job improves confidence in research skills and to assess the quality of research training.

Design and setting  A questionnaire was designed and piloted to assess the content, structure and process of research training.

Population/Sample  All individuals who had passed MRCOG in the five years prior to 1999 and were residents of England or Wales.

Methods  Confidence scores were compared between those with and those without research experience. The availability and value of differing strategies for research training were compared.

Main outcome measures  Confidence in research skills and attitudes to training.

Results  Of the 532 usable questionnaires returned, 226 respondents had done or were doing research and these individuals had higher confidence on a variety of research skills than those with no experience of a formal research job. Confidence was patchy with less than 50% feeling confident at assessing bias in a case-controlled study, understanding the statistics used in a paper or assessing the power of a study. Of those who had done research, 50% or less felt their training had been good or excellent in any area. Self-directed learning and discussion with peers were felt to be the most useful strategies for research training. Short intensive courses were not available for many respondents, but were felt to be useful.

Conclusion  The high levels of dissatisfaction with the training in key skills required for research suggests that there is a need for a system for recognition of research posts. Reform of training in the research job should build on the current strength of encouraging self-directed learning.