In this paper we discuss the detection of CP- and T-violation effects in the framework of a neutrino factory. We introduce three quantities, which are good discriminants for a non-vanishing complex phase (δ) in the 3x3 neutrino mixing matrix: Δ δ , Δ C P and Δ T . We find that these three discriminants (in vacuum) all scale with L/E ν , where L is the baseline and E ν the neutrino energy. Matter effects modify the scaling, but these effects are large enough to spoil the sensitivity only for baselines larger than 5000 km. So, in the hypothesis of constant neutrino factory power (i.e., number of muons inversely proportional to muon energy), the sensitivity on the δ-phase is independent of the baseline chosen. Specially interesting is the direct measurement of T-violation from the ''wrong-sign'' electron channel (i.e., the Δ T discriminant), which involves a comparison of the ν e ->ν μ and ν μ ->ν e oscillation rates. However, the ν μ ->ν e measurement requires magnetic discrimination of the electron charge, experimentally very challenging in a neutrino detector. Since the direction of the electron curvature has to be estimated before the start of the electromagnetic shower, low-energy neutrino beams and hence short baselines, are preferred. In this paper we show, as an example, the exclusion regions in the Δm 2 1 2 -δ plane using the Δ C P and Δ T discriminants for two concrete cases keeping the same L/E ν ratio (730 km/7.5 GeV and 2900 km/30 GeV). We obtain a similar excluded region provided that the electron detection efficiency is ~20% and the charge confusion 0.1%. The Δm 2 1 2 compatible with the LMA solar data can be tested with a flux of 5x10 2 1 muons. We compare these results with the fit of the visible energy distributions.