The identification of winter convection depths from later field observations is an often required but not trivial task. We show from field measurements that the effects of winter convection in the Greenland Sea can vary enormously, including such opposite changes as an increase or decrease of temperature, salinity, or stability. Also, small-scale fluctuations in the vertical can grow larger or cease. Commonly used indicators for past convection in the Greenland Sea like overall vertical homogeneity or a freshwater input to deeper layers alone are therefore not reliable to construct a time series of winter convection. We explain this variety of effects with the occurrence of two rather different ventilation scenarios possible in the Greenland Sea. They are exemplified by two winters, using mainly hydrographic observations of the adjoining summers. The observed differences are substantiated by the results of a one-dimensional model. Out of this a multicriteria catalogue is developed, which can be used to determine unambiguously convection depths in the Greenland Sea. The fulfillment of one of the criteria is sufficient to evidence convection, but the suitable criterion may differ from winter to winter. A straightforward indicator proving the absence of winter convection does not exist.