AbstractLarge-scale simultaneous in vivo recordings of neurons in multiple brain regions raises the question of the probability of recording direct interactions of neurons within, and between, multiple brain regions. In turn, identifying inter-regional communication rules between neurons during behavioural tasks might be possible, assuming conjoint activity between neurons in connected brain regions can be detected. Using the hypergeometric distribution, and employing anatomically-tractable connection mapping between regions, we derive a method to calculate the probability distribution of ‘recordable’ connections between groups of neurons. This mathematically-derived distribution is validated by Monte Carlo simulations of directed graphs representing the underlying anatomical connectivity structure. We apply this method to simulated graphs with multiple neurons, based on counts in rat brain regions, and to connection matrices from the Blue Brain model of the mouse neocortex connectome. Overall, we find low probabilities of simultaneously-recording directly interacting neurons in vivo in anatomically-connected regions with standard (tetrode-based) approaches. We suggest alternative approaches, including new recording technologies and summing neuronal activity over larger scales, offer promise for testing hypothesised interregional communication and source transformation rules.