Novelty and significanceA novel MEA architecture with excellent electrophysiological recordings is presented, where planar microelectrodes are replaced by localized 3 × 3 arrays of mushroom-shaped microstructures. The micro-mushrooms in this islets configuration are not for membrane engulfment but rather for somata entrapment and neurites embracement. As extracellular signals have a significant contribution from axons initial segment, this MEA design also addresses the electrode-neurites electrical coupling. These islets act as strong physical cues, causing topotaxis and increasing by two-fold the probability for somata to localize in the islets. We carry this topotaxis study not only with rat cortical neurons but also with human-derived SH-SY5Y cells.Objective. Planar microelectrode arrays are widely used in neuroscience but have relatively low electrical coupling and signal-to-noise ratio (SNR) in electrophysiology recordings. Strong efforts are therefore being made in improving microelectrode arrays (MEAs) performance, exploring both the microelectrode’s shape and the array’s architecture. Topographical features can be used in MEAs for promoting neuron-microelectrode contact, making 3D-microstructured MEAs an interesting design strategy for better electrophysiology measurements. Approach. Here, we present a novel MEA architecture, where planar microelectrodes are replaced by localized 3 × 3 arrays of mushroom-shaped microstructures. Contrarily to previous studies, the purpose for the micro-mushrooms in this islets configuration is not membrane engulfment but rather entrapment, for somata, and embracement, for neurites. Main results. We show that these islet-like agglomerates of micro-mushrooms act as strong physical cues, causing topotaxis and increasing the probability by two-fold for somata to localize in the islets, and neurites to curl on the microelectrodes. Importantly, we carry this topotaxis study not only with rat cortical neurons but also with human-derived SH-SY5Y cells. With recent evidence that extracellular signals have a significant contribution from axons initial segment it becomes clear that MEA designs should also address the electrode-neurites coupling. We detail the fabrication process of these chips, designed to be compatible with a standard MEA recording system, and make the computer-aided design (CAD) publically available. We also demonstrate the electrophysiological capabilities of this new MEA by electrochemical impedance spectroscopy and recordings of cortical and hippocampal neurons, showing excellent SNR. Significance. Overall this new MEA islets configuration has a significant impact in the array efficiency and contributes towards improved high yield and high fidelity/quality extracellular recordings from mammalian neurons.