We consider energy-efficient adaptive power allocation for three incremental multiple-input multiple-output (IMIMO) systems employing ARQ, hybrid ARQ (HARQ) with Chase combining (CC), and HARQ with incremental redundancy (IR), to minimize their rate-outage probability (or equivalently packet drop rate) under a constraint on average energy consumption per data packet. We first provide the rate-outage probability expressions for the three IMIMO systems, and then propose methods to convert them into a tractable form and formulate the corresponding non-convex optimization problems that can be solved by an interior-point algorithm for finding a local optimum. To further reduce the solution complexity, using an asymptotically equivalent approximation of the rate-outage probability expressions, we approximate the non-convex optimization problems as a unified geometric programming problem (GPP), for which we derive the closed-form solution. Illustrative results indicate that the proposed power allocation (PPA) offers significant gains in energy savings as compared to the equal-power allocation (EPA), and the simple closed-form GPP solution can provide closer performance to the exact method at lower values of rate-outage probability, for the three IMIMO systems.