In this type of study, a defined population sample is followed forward in time. Variables of interest are assessed at baseline, before development of the disease. At the end of the follow-up period, baseline measurements are compared between those who went on to develop the disease and those who remained free of the disease. This type of study is a much more powerful tool than the case-control study as cases are assessed before the onset of disease. Thus, it does not suffer the problem of assessment of the past. It is, however, much more costly in terms of the resources required. This is because a large population sample and a lengthy follow-up period are usually needed. A prospective study of cancer, for example, would require baseline measurements to be made on several thousand individuals and these would need to be followed up for about 20 years in order to yield a sufficient number of new cases of the disease.

Subjects selected for the study should be representative of the kind of people to whom the hypothesis relates. Subjects should be followed up as completely as possible - those lost to follow-up are likely to differ in many ways from those who can be traced. If a substantial proportion of subjects are not followed up, then any conclusions from the study may be misleading. One difficulty is that subjects may move during the follow-up period, making them difficult to trace. Some prospective studies have used specific occupational groups such as doctors or civil servants as these are relatively easy to follow-up. If using the general population, subjects' National Health Service numbers and National Insurance numbers can be useful in tracing them.

Prospective studies have a number of other difficulties. For example, some of the baseline measurements may change over time. For example, people may reduce their fat intake or increase their fibre intake; some vegetarians may lapse, whereas some non-vegetarians may become vegetarian; some changes may occur in medical treatment during the follow-up period. Such changes may reduce the chance of detecting an association between these variables and the disease of interest. A further problem is that since the study is observational, any associations detected may or may not be causal.

If subjects are seen again at intervals during the follow-up period, additional measurements can be made. This helps in establishing how consistent measurements made at baseline are and improves the ability to interpret the study results. The number of subjects selected for the study and the duration of follow-up needs to be carefully calculated so as to ensure a sufficient number of new cases of the disease in order to detect differences between new cases and controls that are likely to be of clinical importance.