Here we analyze a sample of close galaxy pairs (relative projected separation < 25 h⁻¹ kpc and relative radial velocities < 350 km s⁻¹) using a weak-lensing analysis based on the Canada-France-Hawaii Telescope Stripe 82 Survey (CS82). We determine halo masses for the total sample of pairs as well as for interacting, red, and higher-luminosity pair subsamples with ∼3σ confidence. The derived lensing signal for the total sample can be fitted either by a Singular Isothermal Sphere (SIS) with σ_V = 223 ± 24 km s⁻¹ or a Navarro–Frenk–White (NFW) profile with R₂₀₀ = 0.30 ± 0.03 h⁻¹ Mpc. The pair total masses and total r band luminosities imply an average mass-to-light ratio of ∼200 h M_⊙/L_⊙. On the other hand, red pairs which include a larger fraction of elliptical galaxies, show a larger mass-to-light ratio of ∼345 h M_⊙/L_⊙. Derived lensing masses were compared to a proxy of the dynamical mass, obtaining a good correlation. However, there is a large discrepancy between lensing masses and the dynamical mass estimates, which could be accounted for by astrophysical processes such as dynamical friction, by the inclusion of unbound pairs, and by significant deviations of the density distribution from SIS and NFW profiles in the inner regions. We also compared lensing masses with group mass estimates, finding very good agreement with the sample of groups with two members. Red and blue pairs show large differences between group and lensing masses, which is likely due to the single mass-to-light ratio adopted to compute the group masses.