The processes involved in the shock initiation of detonation in nitromethane (NM) have been the subject of many experimental and theoretical studies. These generally support the classical homogeneous model though some details of the build-up process are still controversial. In order to clarify these points, we have performed plate impact experiments to study the initiation of NM detonation under conditions of steady one dimensional strain, for shock pressures ranging from 8.5 to 12 GPa. For each shock pressure we have studied the influence of i) the roughness of the plate in contact with the explosive and ii) the natural gaseous micro-bubbles present in NM. A six wavelength optical pyrometer, with 3 ns rise-time and a temperature range of 1500-6000 K, was used to determine the temperature during shock-to-detonation transition (SDT). A Fabry-Perot interferometer, a capacitor transducer and piezoelectric pins were also used to determine the sequence of events during the initiation process and to contribute to the analysis of the temperature profiles. The experimental results confirm the main steps of the classical homogeneous model, but also show that the build-up process is more complex than the picture given by Chaiken. Unlike Chaiken assumptions, some chemical reactions occur near the plate/NM interface before the establishment of the superdetonation, which seems to be initiated inside the NM, at a run distance from the barrier that depends on the shock level.