Conventional approaches to accelerated simultaneous mul-tislice (SMS) MRI rely on structured k-space sampling and parallel imaging with known coil sensitivity profiles. In this paper, we introduce a novel framework for SMS MRI that is flexible enough to accommodate a number of different experimental variations: it supports both single-channel and parallel imaging data, both calibration-based and calibration-less k-space sampling trajectories, and Hadamard, Fourier, and random-phase non-Fourier encoding along the slice dimension. Our proposed SMS framework is based on the recently introduced LORAKS framework (low-rank matrix modeling of local k-space neighborhoods). The new framework, which we call SMS-LORAKS, is evaluated using real retrospectively undersampled k-space data. These evaluations confirm the promise and flexibility of the proposed approach.