Phasor measurement units (PMUs) are the measurement devices fostering the transformation of electric power networks towards the smart grid paradigm. They should accurately measure synchrophasors, frequency, and rate of change of frequency (ROCOF), so that the management and control applications relying on PMU-based distributed monitoring systems can operate effectively. Commercial PMUs performance is typically guaranteed by the compliance with the IEEE standard C37.118.1, which is focused on PMUs for power transmission systems and defines testing conditions and error limits. However, actual operating conditions are much more variable than those covered by the standard, especially when PMUs are used in distribution networks. In particular, the standard does not consider unbalance, which may be negligible neither in transmission nor in distribution grids. For the first time, this paper analyzes the impact of unbalance on the accuracy of four of the most significant classes of signal processing algorithms for PMU measurements. Synchrophasor, frequency, and ROCOF estimation performances under different unbalance conditions are investigated in the test cases suggested by the IEEE C37.242-2013 guide. Novel analytic expressions to predict the errors are derived and validated, and they are proved to be useful for an effective implementation of PMU algorithms intended for both distribution and transmission systems.