Modeling sanitary sewer overflow (SSO) discharges based on flows measured at pumping stations presents specific challenges distinct from modeling combined sewer overflow (CSO) discharges and is insufficiently studied. This paper presents a methodology aimed at reducing and assessing the uncertainty in SSO estimation. Four lumped models are compared: a time–area curve, linear reservoir, Clark, and simplified numerical modeling using SWMM. Clark’s model delivers the best results, with very reasonable estimates. Although the simplified SWMM model also provides good estimates, factors such as determinism in weir crest height and potential storage in the sewer network introduce additional uncertainties and errors. The attenuation of measured hydrographs at pumping mains leads to an underestimation of both the catchment area and the SSO discharges, which depends on the pump operating frequency and requires further evaluation. The maximum flow rate threshold allowed for the pumping station is a key parameter for SSO frequency and volume; its influence is evaluated in detail. This methodology can also be used to assess unknown overflows upstream of monitored weirs, as it provides estimates of total SSO discharges occurring in a catchment. The results of applying this methodology to two contrasting and paradigmatic pumping stations are discussed.