Most stellar flares’ soft X-ray light curves possess a ‘typical’ morphology, which consists of a rapid rise followed by a slow exponential decay. However, a study of 216 of the brightest flares on 161 pre-main-sequence stars, observed during the Chandra Orion Ultradeep Project (COUP), showed that many flare light curves depart from this typical morphology. While this can be attributed to the superposition of multiple typical flares, we explore the possibility that the time-variable eclipsing of flares by their host stars may also be an important factor. We assume each flare is contained within a single, uniform plasma density magnetic loop and specify the intrinsic variation of the flare’s emission measure with time. We consider rotational eclipse not only by the star itself, but also by circumstellar discs and flare-associated prominences. Based on this simple model, we generate a set of flares similar to those observed in the COUP data base. Many eclipses simply reduce the flare’s maximum emission measure or decay time. We conclude therefore that eclipses often pass undetected, but usually have only a modest influence on the flare emission measure profile and hence the derived loop lengths. We show that eclipsing can easily reproduce the observed atypical flare morphologies. The number of atypical modelled flare morphologies is, however, much less than that found in the COUP sample. The large number of observed atypical flare morphologies, therefore, must be attributed to other processes such as multiple flaring loops.