This book investigates reliability and security challenges in neuromorphic computing systems based on memristors. Conventional computing faces performance limits due to the separation of memory and processing, known as the von Neumann bottleneck. Computing-in-memory (CIM) addresses this by performing operations inside memory, either using analog CIM for high throughput or logic-in-memory (LIM) for simpler digital operations. Memristors enable these paradigms due to their high density, non-volatility, and low power. However, device variability and aging introduce reliability issues, while hardware security vulnerabilities threaten system integrity. This work introduces a fault injection framework, including X-Fault and FLIM, to study LIM reliability from device to application levels. It also presents NeuroHammer, a novel hardware attack that exploits memristor properties to induce faults and compromise system security. Finally, the NeuroBreakoutBoard (NBB) platform is developed to precisely characterize memristive devices and support CIM operations. Together, these contributions aim to improve the dependability and security of memristor-based computing systems.