This thesis concerns the development of numerical macromodels of digital integrated circuits input/output buffers using system identification methods. Chapter 1 introduces the need for accurate and efficient macromodels of active devices for system-level simulation of signal integrity and electromagnetic compatibility effects in electronic equipment. Chapter 2 reviews existing modeling approaches for integrated circuits. Chapter 3 discusses the parametric modeling process including model selection, identification signals, parameter estimation, validation and implementation. Chapter 4 presents different parametric model representations considered: Sigmoidal Basis Functions, Echo State Networks, and Local Linear State-Space models. Chapter 5 assesses the performance of these models on a test device, showing Local Linear State-Space models provide the best results. Chapter 6 applies the modeling process