A growing research interest exists in the field of modular quadrotor platforms as well as in load carrying capabilities of aerial robots. This inevitably leads to the problem of changing system parameters that consequently cannot be considered available for control design. In this article, two variants of adaptive state space controllers for attitude stabilisation and self-tuning of a quadrotor are proposed that do not depend on any knowledge of the inertia tensor. First of all, the use of a model identification adaptive controller (MIAC) is suggested in terms of combining a recursive least-squares estimator with exponential forgetting with an integral discrete-time state space controller. Furthermore, a continuous-time model reference adaptive control (MRAC) scheme based on Lyapunov theory is applied. The effectiveness of the adaptive methods is demonstrated in simulations with a quaternion-based non-linear dynamic model of a quadrotor derived in this work and compared to a non-adaptive integral state space controller.