In clinical and epidemiological studies, competing risks data arise when the subject can experience one, and only one, of several mutually exclusive types of events. Competing risks data are often right- or interval-censored. For right-censored data, a semiparametric regression model proposed by Fine and Gray (1999) has become the method of choice for formulating the effects of covariates on the cumulative incidence. Its estimation, however, requires modeling of the censoring distribution and is not statistically efficient. In this project, we present a broad class of semiparametric transformation models which extends the Fine and Gray model, and we derive the nonparametric maximum likelihood estimators (NPMLEs). We develop a simple and fast algorithm for computing the NPMLEs through the profile likelihood. We establish the consistency, asymptotic normality, and semiparametric efficiency of the NPMLEs. In addition, we construct graphical and numerical procedures to evaluate and select models. Then, we demonstrate the advantages of the proposed methods over the existing ones through extensive simulation studies and an application to a major study on bone marrow transplantation. We extend the same class of transformation models to interval-censored competing risks data. We allow covariates to be time-dependent and accommodate missing event type information. We develop a novel EM algorithm to compute the NPMLEs, and establish the consistency, asymptotic normality, and semiparametric efficiency of the NPMLEs. Extensive numerical studies show that our methods perform well in finite samples. A well-known HIV/AIDS study is provided to illustrate our methods. Finally, we consider two problems which can be viewed as extensions of the methodologies described. One is for partly interval-censored competing risks data, where some of the risks are interval censored while the rest are right censored. The other concerns interval-censored failure time with a continuous mark. We describe semiparametric regression methods for these data types and analyze data from HIV/AIDS studies using the proposed procedures.