Correlated multiple sampling differential averaging (CMS-DA) is useful for realizing low-noise column readout circuits in CMOS image sensors. CMS-DA involves sampling the signal and reset levels multiple times and differentiating the averages. In other words, this method is like averaging of correlated double sampling (CDS) for the multiple sampling of signal and reset levels. This paper analyzes the effectiveness of CMS-DA for reducing 1/f noise and the analysis is in good agreement with time-domain simulations.