Abstract The high-cadence survey of the Zwicky Transient Facility (ZTF) has completely dominated the discovery of tidal disruption events (TDEs) in the past few years and resulted in the largest sample of TDEs with optical/UV light curves well sampled around their peaks, providing us an excellent opportunity to construct a peak luminosity function (LF) of tidal disruption flares (TDFs). The new construction is necessary particularly considering that the most updated LF reported in literature has been inferred from only 13 sources from five different surveys. Here we present the optical and blackbody LFs calculated by 33 TDFs discovered in the ZTF-I survey. The optical LF can be described by both a power-law profile dN / dL g ∝ L g − 2.3 ± 0.2 , and a Schechter-like function. The blackbody LF can be described by a power-law profile dN / dL bb ∝ L bb − 2.2 ± 0.2 , shallower than the LF made of the previous van Velzen (2018) sample. A possible low-luminosity turnover in the optical LF supports an Eddington-limited emission scenario. The drop of the volumetric rate at high luminosity suggests a rate suppression due to direct captures of the black hole. The total volumetric rate is 1 order of magnitude lower than the previous estimation, which is probably not simply caused by the high fraction postpeak sources (7/13) in the previous sample. Instead, the normalization step during the previous LF construction to reconcile various surveys might adversely amplify the influence of serendipitous discoveries. Therefore, TDFs selected from ongoing and upcoming uniform surveys like ZTF, Vera Rubin Observatory, and the Wide-Field Survey Telescope should yield more accurate LFs.