We present the first direct comparison between Balmer line and panchromatic spectral energy distribution (SED)-based star formation rates (SFRs) for z ∼ 2 galaxies. For this comparison, we used 17 star-forming galaxies selected from the MOSFIRE Deep Evolution Field (MOSDEF) survey, with 3σ detections for Hα and at least two IR bands (Spitzer/MIPS 24 μm and Herschel/PACS 100 and 160 μm, and in some cases Herschel/SPIRE 250, 350, and 500 μm). The galaxies have total IR (8-1000 μm) luminosities of ∼ 1011.4-1012.4 L⊙ and SFRs of ∼30-250 M⊙ yr-1. We fit the UV-to-far-IR SEDs with flexible stellar population synthesis (FSPS) models - which include both stellar and dust emission - and compare the inferred SFRs with the SFR(Hα, Hβ) values corrected for dust attenuation using Balmer decrements. The two SFRs agree with a scatter of 0.17 dex. Our results imply that the Balmer decrement accurately predicts the obscuration of the nebular lines and can be used to robustly calculate SFRs for star-forming galaxies at z ∼ 2 with SFRs up to ∼ 200 M⊙ yr-1. We also use our data to assess SFR indicators based on modeling the UV-to-mid-IR SEDs or by adding SFR(UV) and SFR(IR), for which the latter is based on the mid-IR only or on the full IR SED. All these SFRs show a poorer agreement with SFR(Hα, Hβ) and in some cases large systematic biases are observed. Finally, we show that the SFR and dust attenuation derived from the UV-to-near-IR SED alone are unbiased when assuming a delayed exponentially declining star formation history.