In the present analysis experimental results and computer modelling have been used to compare biomechanics of the two most contrasting lantern types among regular echinoids, i.e. the cidarid-type (Stylocidaris affinis) and the camarodont (Paracentrotus lividus) lantern. The lantern is modelled as a rigid pyramid anchored in the centre of the peristomial area. Each side is connected to the test by means of muscles (protractors and retractors) rind ligaments (compass depressors). The overall movement of the lantern may be resolved into vertical displacement and lateral tilting. Computer-aided mechanical analysis of the system takes into account the action of protractor and/or retractor muscles, their elastic reaction and the interaction with the ligamentous structures (peristomial membrane and compass depressors). The geometrical parameters have been determined by preliminary structural analysis and the values of active and passive forces have been measured experimentally. The results of simulations highlight important differences in the mechanics of the Stylocidaris lantern compared to that of Paracentrotus: (1) a very much greater antagonistic effect of muscular passive stretch resistance on lateral tilting: (2) a minor role of the peristomial membrane tensile strength and muscular activity in controlling vertical lantern displacement: (3) a less efficient lantern retraction and protraction owing to unfavourable muscle insertion position on the perignathic girdle and ineffective muscular contraction forces. As a conclusion, in Stylocidaris the action of the protractor and retractor muscles seems to play no significant role in moving the overall lantern, their action being mainly exerted to stabilize it. On the contrary, in Paracentrotus such stabilizing action is provided by the compass depressor ligaments. Finally, on theoretical grounds the arrangement of muscle attachments around the perignathic girdle in Stylocidaris seems to be just as good as that of Paracentrotus for controlling overall lantern mobility, if relative heights of insertions are ignored.