First results of a high precision mass measurement program for very short-lived nuclides C. Toader*, G. Audi, G. Conreur, H. Doubre, S. Henry, M. Jacotin, J.F. Képinski, G. Le Scornet, D. Lunney, C. Monsanglant, M. de Saint Simon, C. Thibault CSNSM-IN2P3-CNRS, Bat 108, F-91405 Orsay-campus C. Borcea, M. Duma INPE, P.O. Box MG-6, RO-76900 Bucharest-Magurele G.Lebee CERN, CH-1211 Geneva 23 H-J. Kluge GSI, Planckstrasse 1, D-64291 Darmstadt G. Bollen Univ. Munich, Coulombwall 1, D-85748 Garching and the ISOLDE collaboration CERN, CH-1211 Geneva 23 Recent commissioning of the new experimental program MISTRAL (Mass Measurements at ISolde using a Transmission and Radiofrequency spectrometer on Line) brought mass measurements of high precision performed on sodium, magnesium, aluminum, potassium, calcium and titanium isotopes. This spectrometer is using a technique of radiofrequency excitation of ion trajectories in a homogeneous magnetic field. The mass is obtained by the comparison of the cyclotron frequency of the unknown mass to that of a reference mass. This measurement procedure is well suited for direct mass measurements of very short-lived nuclides, the shortest half-life of an isotope measured with MISTRAL was 30.5 ms (28Na). Though this first period of data taking was dedicated to the exploration of MISTRAL performances it has allowed us to reduce the uncertainty in the mass of the most exotic isotopes of sodium by almost one order of magnitude. Even for the weakly produced 30Na (T1/2=48ms), the accuracy was still 8*10^-7. Verifying binding energies and minimizing their uncertainties in this region of the nuclear chart is important to clarify the long standing problem of the strength of the N=20 shell closure. After an introduction concerning the principles of MISTRAL we will explain the handling of the systematic errors and we will discuss the results obtained for sodium isotopes. * Now at GSI-Darmstadt