We present here a joint-inversion algorithm to build a resistivity, P-wave, and S-wave velocity model from apparent resistivity, surface wave dispersion and P-wave refraction data. This algorithm can also include a-priori information available for the site, as well as any physical links among the model parameters, and the result is an internally consistent multi-parametric model. The obtained model resolves more properly the true model because the joint inversion mitigates some problems related to the individual inversion of each type of experimental data like solution non-uniqueness, illness, or lack of resolution, which might lead to interpretation ambiguities. We describe the proposed algorithm and we show the result of its application on a smoothly laterally varying synthetic model.