Anglican parish registers have been the basis for most studies of population trends and characteristics in early modern England, and one of the most important of the techniques used in analysing them has been family reconstitution. But Anglican registers at all times were an incomplete and inaccurate record of vital events, and their defects tended to become more pronounced in the later part of the period during which they afford the chief source of empirical information about population behaviour. And there are inherent limitations and biases in the results that can be obtained by family reconstitution. This article attempts to describe the range of difficulties and dilemmas involved in studying the demography of populations in the past when using this source of data and this technique of analysis. A variety of tests is deployed to establish the degree of reliability attaching to the results obtained in a recent exercise based on the family reconstitution of 26 parishes, and more generally to assess the opportunities open to scholarship in this area and the pitfalls associated with such work. The conclusion is that reliable results can be obtained but that great care is needed in the selection of suitable registers, and that a number of tests should be employed to monitor the internal consistency and the demographic plausibility of any findings.