This chapter concludes the book by summarizing the main findings on reliability and security in memristor-based neuromorphic computing. A fault injection platform was introduced to analyze the resilience of logic-in-memory operations under realistic fault conditions. Results showed that memristive systems can tolerate limited fault levels, with stuck-at faults having a stronger impact than bit-flip faults. The work also revealed hardware security threats arising from intrinsic device properties. NeuroHammer demonstrated that thermal crosstalk in dense arrays can induce controlled bit flips, exposing a significant attack surface. To complement simulations, the NeuroBreakoutBoard platform was presented, enabling device- and array-level characterization as well as execution of computing-in-memory operations on physical crossbars. From these results, three requirements emerge for dependable neuromorphic platforms: reliability must be addressed across the stack, application behavior must be considered in fault management, and hardware security must be integrated early in design. Promising research directions are outlined.