Although Excel contains only a handful of (inverse cumulative) probability distribution functions that allow distributions to be sampled directly, in fact it is easy to create samples of around 20 distributions using simple Excel formulae and (simple) VBA code in some cases.
These topics are discussed in detail in Chapter 10 of my recent book Business Risk and Simulation Modelling in Practice (see About Business Risk and Simulation Modelling in Practice). In fact (and fortunately) the in-built Excel functions correspond to those distributions where no simple analytic solution exists, whereas for the other functions, either simple analytic solutions exist, or iterative methods are required.
For example, Excel contains functions such as NORMSINV (NORMS.INV is more recent versions) which allows formulae such as NORMSINV(RAND()) [when written in generic short form for simplicity] to be used to create sample from a Normal distribution. On the other hand, there is no inverse function for a Weibull distribution in Excel, whereas the inverse of a Weibull distribution can be written quite simply (as a function of a RAND() function to represent the probability and the two parameters for a Weibull distribution). Similarly, the inverse function for an alternative (percentile) parameterisation of the distribution is easy to determine analytically, so that the sample is a function of the given percentiles (say P10 and P90) as well as the RAND() sample. The required formulae for these are given in the book, or can be derived (fairly) easily from the mathematical equations for the distributions.