The article reviews recent advances in information theory that govern the transmission of short packets, which are crucial for supporting novel traffic types in upcoming wireless systems, particularly in 5G. Short packets are essential for Machine-to-Machine (M2M) communications and critical applications requiring low latency and ultra-high reliability. Current wireless systems are not designed to handle short-packet transmissions, as metadata is often assumed to be negligible compared to the payload. However, in short-packet communications, metadata can be of similar size to the payload, necessitating more efficient encoding methods. The authors introduce the concept of the maximum coding rate at finite packet length and error probability, which is more relevant for short packets. They apply this metric to three scenarios—two-way channel, downlink broadcast channel, and uplink random access channel—to illustrate how control information can be optimized in short-packet communications. The insights from these examples suggest that new principles are needed for the design of wireless protocols supporting short packets, which will have a direct impact on system design. The paper also discusses the challenges in designing protocols for ultra-reliable communication (URC) and massive machine-to-machine communications (MM2M) applications, emphasizing the need to address the trade-offs between reliability, throughput, and PHY overhead. The authors provide a comprehensive review of information-theoretic performance metrics and their application to short-packet communications, highlighting the limitations of traditional metrics like capacity and outage capacity when applied to short packets. They introduce the maximum coding rate as a more accurate measure for assessing the performance of short-packet communications, especially in the context of delay-constrained applications.The article reviews recent advances in information theory that govern the transmission of short packets, which are crucial for supporting novel traffic types in upcoming wireless systems, particularly in 5G. Short packets are essential for Machine-to-Machine (M2M) communications and critical applications requiring low latency and ultra-high reliability. Current wireless systems are not designed to handle short-packet transmissions, as metadata is often assumed to be negligible compared to the payload. However, in short-packet communications, metadata can be of similar size to the payload, necessitating more efficient encoding methods. The authors introduce the concept of the maximum coding rate at finite packet length and error probability, which is more relevant for short packets. They apply this metric to three scenarios—two-way channel, downlink broadcast channel, and uplink random access channel—to illustrate how control information can be optimized in short-packet communications. The insights from these examples suggest that new principles are needed for the design of wireless protocols supporting short packets, which will have a direct impact on system design. The paper also discusses the challenges in designing protocols for ultra-reliable communication (URC) and massive machine-to-machine communications (MM2M) applications, emphasizing the need to address the trade-offs between reliability, throughput, and PHY overhead. The authors provide a comprehensive review of information-theoretic performance metrics and their application to short-packet communications, highlighting the limitations of traditional metrics like capacity and outage capacity when applied to short packets. They introduce the maximum coding rate as a more accurate measure for assessing the performance of short-packet communications, especially in the context of delay-constrained applications.