5G networks are expected to be used in many markets, one of which is the Factories of the Future (FoF). In the FoF, applications like regular monitoring and controlling of components (e.g. temperature) are introducing the need of massive deployment of sensors and actuators. Additionally, sensors which are monitoring time-critical components of the factory (e.g. pressure in power plants) should experience low delays with a high reliability. The current LTE-based technology for machine-type communications, namely Cat-M1, imposes limitations in supporting massive connectivity and application with low delay and high reliability requirements, primarily due to the so-called ‘Random Access’ (RA) procedure which is used by the devices to establish a connection with the base station, before the actual transmission of their data. The key objectives of this study are to assess the RA procedure currently used in Cat-M1 networks as a baseline, and furthermore design and evaluate a new RA procedure for 5G networks, which targets typical FoF applications and their requirements. Based on the simulation assessment of the procedures studied, it was found that the so-called ‘Two-Step RA procedure’ is performing the best regarding end-to-end delay. Specifically, it was found that e.g. in a FoF network with 3000 devices an end-to-end delay of 16 ms can be achieved with 99.99% reliability, introducing a gain of 64 ms compared to the end-to-end delay in Cat-M1 networks.