We analyze the impact of recent solar and atmospheric data in the determination of the neutrino oscillation parameters, taking into account that both the solar ν e and the atmospheric ν µ may convert to a mixture of active and sterile neutrinos. We use the most recent global solar neutrino data, including the 1496-day Super-K neutrino data sample, and we investigate in detail the impact of the recent SNO neutral current, spectral and day/night data by performing also an analysis using only the charged current rate from SNO. We confirm the clear preference of the pure active LMA solution of the solar neutrino problem and obtain that the LOW, VAC, SMA and Just-So 2 solutions are disfavored with a ∆χ 2 = 9, 9, 23, 31, respectively. Furthermore, we find that the global solar data constrains the admixture of a sterile neutrino to be less than 44% at 99% C.L..A pure sterile solution is ruled out with respect to the active one at 99.997% C.L.. By performing an improved fit of the atmospheric data, we also update the corresponding regions of oscillation parameters. We find that the recent atmospheric Super-K (1489-day) and MACRO data have a strong impact on constraining a sterile component in atmospheric oscillations: if the ν µ is restricted to the atmospheric mass states only a sterile admixture of 16% is allowed at 99% C.L., while a bound of 35% is obtained in the unconstrained case. Pure sterile oscillations are disfavored with a ∆χ 2 = 34.6 compared to the pure active case.In the appendix we discuss the implications of the first 145.1 days of KamLAND data on the determination of the solar neutrino parameters. The inclusion of KamLAND enhances the rejection of non-LMA-MSW solutions by 13 units in ∆χ 2 . The bound on the sterile neutrino fraction is practically unaffected in the boron-fixed case, while it improves from 61% to 51% in the boron-free case.