Circadian clocks regulate the timing of various behavioral and physiological activities in most organisms on a 24-hr scale such that they are phased appropriately to external, cyclic changes in the environment. The clock neuronal network in Drosophila melanogaster is comprised of  ~150 neurons distributed bilaterally in the brain, which regulate rhythmic behaviors. A subset of these neurons, the small ventral lateral neurons (s-LNvs) also known as the M-cells, secrete the neuropeptide Pigment Dispersing Factor (PDF) which functions as a synchronizing factor in the network with complex effects on other clock neurons. Lack of PDF and its receptor, PDFR, result in most flies displaying arrhythmicity in activity-rest cycles under constant conditions. However, as is the case in most studies of neuronal circuits that control behavior, the majority of the work has been conducted in males. Our results show that some assumptions that had been made for the male circadian network are not true for females. We report that the female circadian rhythms are less affected by the loss of both PDF and PDFR. Cell-specific Pdf knockout has a greater effect on male rhythms in fundamental circadian properties like free-running period and rhythm power. In addition, adult specific manipulation of Pdf results in changes in rhythmic power, but no changes in free-running period, suggesting that this neuropeptide plays important functions during the development of the circadian timekeeping network. Finally, we tested the influence of the M-cells over the circadian and showed that speeding up the molecular clock specifically in M cells leads to sexually dimorphic phenotypes. Our results suggest that the female circadian system is more resilient to the loss of the synchronizing factor PDF, and that circadian timekeeping is more distributed across the clock neuron network in females. Taken together, our data suggest that the relative influence of different subsets of neuronal circadian oscillators is sexually dimorphic and that anatomically similar networks can have functional differences in generating behavioral outputs.