Explaining macroevolutionary divergence in light of population genetics requires understanding the extent to which the patterns of mutational input contribute to long-term trends. In the context of quantitative traits, mutational input is typically described by the mutational variance-covariance matrix, or the M-matrix, which summarizes phenotypic variances and covariances introduced by new mutations per generation. However, as a summary statistic, the M-matrix does not fully capture the pleiotropic structure of the underlying mutational architecture, and there exist infinitely many possible underlying mutational architectures that give rise to the same M-matrix. Here, using simulations, we demonstrate alternative mutational architectures underlying the same M-matrix can lead to different levels of constraint on evolution and result in difference in rate and dynamics of adaptive evolution. We also found for a given M-matrix, the evolutionary variance-covariance matrix, or the R-matrix, is more strongly aligned to the M-matrix when the underlying mutational architecture has stronger pleiotropy. Taken together, our results reveal that aspects of mutational input not reflected by the M-matrix can have a profound impact on long-term evolution, and suggest it is important to take them into account in order to connect macroevolutionary patterns to microevolutionary mechanisms.