The well-known model of Tate and Alekseevskii for projectile penetration seems to be incapable of simultaneous prediction of both penetration depth and remaining projectile length. The latter is calculated to be negligible for relevant combination of projectile and target materials. Experimental experience is instead that the remaining projectile length typically may be about half the penetration depth and that there is only negligible loss of projectile material. Therefore a quite new model for rigid, deformable and consumable projectile penetration is suggested, which appears to predict experimental results with sufficient accuracy. For deformable projectile penetration the model makes use of a friction coefficient between the projectile and the target, the numerical value of which is found to be 0.1. The model first accounts for the initial penetration phase, when part of the projectile has not yet passed the impacted surface of the target. Thereupon the continued penetration is calculated with appropriate modification of the retardation forces. All results are obtained in analytic form. The analytic relations implicitly give certain results, which are easy to calculate numerically. It seems as if the models for rigid and deformable projectile penetration should be able to account for almost all cases of practical interest in connection with assessments of effects and vulnerability of complex targets, which is the purpose of the presented work.