We will discuss the following general question: Is it possible to provide an upper pointwise bound for a given probability density by means of its Laplace transform? Although in general this is not possible, the answer turns out to be affirmative when the density has a convolution structure (that is, the convolution of two or more densities).