We tackle the problem of estimation level sets Lf(λ) density f a random vector X supported on smooth manifold M⊂Rd, from an iid sample X. To do that we introduce kernel-based estimator fˆn,h, which is slightly modified version one proposed in Rodríguez-Casal and Saavedra-Nieves (2014) proves its a.s. uniform convergence to f. Then, propose two estimators Lf(λ), first plug-in: Lfˆn,h(λ), proven ...