The column collapse experiment is a simplified version of the complex granular flows observed in both natural and industrial contexts. Due to its simple setup and rich behavior, the column collapse has been studied experimentally and numerically by several authors. The purpose of the analyses presented in this paper is to verify whether some of the results presented in these publications are affected by grain-size effects. In order to do so, we simulate two-dimensional granular columns by means of a discrete-element method, i.e., contact dynamics. Specifically, we study the influence of the grain size, as compared to the system size, on the resultant deposit geometry and collapse duration. We show that (i) some of the previously published results may be affected by grain-size effects, (ii) in order to avoid these effects, the system-size to grain-size ratio must be larger than 75 for short columns and larger than 50 for tall columns, and (iii) the quantities that are the most affected by grain-size effects are the column mobility and the collapse duration. Our findings serve as a tool for comparing results obtained by different researchers and draw guidelines on the number of grains that must be used in order to avoid grain-size effects.