A number of stability criteria exist for dark energy theories, associated with requiring the absence of ghost, gradient and tachyonic instabilities. Tachyonic instabilities are the least well explored of these in the dark energy context and we here discuss and derive criteria for their presence and size in detail. Our findings suggest that, while the absence of ghost and gradient instabilities is indeed essential for physically viable models and so priors associated with the absence of such instabilities significantly increase the efficiency of parameter estimations without introducing unphysical biases, this is not the case for tachyonic instabilities. Even strong such instabilities can be present without spoiling the cosmological validity of the underlying models. Therefore, we caution against using exclusion priors based on requiring the absence of (strong) tachyonic instabilities in deriving cosmological parameter constraints. We illustrate this by explicitly computing such constraints within the context of Horndeski theories, while quantifying the size and effect of related tachyonic instabilities.