Abstract  This review focuses on how smooth muscle sarcoplasmic reticulum (SR), the major releasable Ca2+ store in these cells, performs its many functions by communicating with the plasma membrane (PM) and other organelles across cytoplasmic nanospaces, defined by membrane–membrane junctions less than 50 nm across. In spite of accumulating evidence in favour of the view that cytoplasmic nanospaces are a prerequisite for effective control of diverse cellular functions, our current understanding of how smooth muscle cells accomplish site- and function-specific Ca2+ signalling remains in its infancy. We first present evidence in support of the view that effective Ca2+ signalling depends on the restricted diffusion of Ca2+ within cytoplasmic nanospaces. We then develop an evidence-based model of the smooth muscle SR – the ‘pan-junctional SR’ model – that incorporates a network of tubules and quilts that are capable of auto-regulating their Ca2+ content and determining junctional [Ca2+]i through loading and unloading at membrane–membrane nanojunctions. Thereby, we provide a novel working hypothesis in order to inform future investigation into the control of a variety of cellular functions by local Ca2+ signals at junctional nanospaces, from contraction and energy metabolism to nuclear transcription. Based on the current literature, we discuss the molecular mechanisms whereby the SR mediates these multiple functions through the interaction of ion channels and pumps embedded in apposing membranes within inter-organellar junctions. We finally highlight the fact that although most current hypotheses are qualitatively supported by experimental data, solid quantitative simulations are seriously lacking. Considering that at physiological concentrations the number of calcium ions in a typical junctional nanospace between the PM and SR is of the order of 1, ion concentration variability plays a major role as the currency of information transfer and stochastic quantitative modelling will be required to both test and develop working hypotheses.