The Uranian satellite Miranda presents a high inclination (4.338 • ) and evidences of resurfacing. It is accepted since 20 years (e.g. Wisdom 1989, Malhotra and Dermott 1990) that this inclination is due to the past trapping into the 3:1 resonance with Umbriel. These last years there is a renewal of interest for the Uranian system since the Hubble Space Telescope permitted the detection of an inner system of rings and small embedded satellites, their dynamics being of course ruled by the main satellites. For this reason, we here propose to revisit the long-term dynamics of Miranda, using modern tools like intensive computing facilities and new chaos indicators (MEGNO and frequency map analysis). As in the previous studies, we find the resonance responsible for the inclination of Miranda and the secondary resonances associated, likely to have stopped the rise of Miranda's inclination at 4.5 • . Moreover, we get other trajectories in which this inclination reaches 7 • . We also propose an analytical study of the secondary resonances associated, based on the study by Moons and Henrard (1993).