The identification of tumor-associated antigens and new insights into the molecular and cellular basis of tumor immunology have generated a variety of vaccine strategies. Although promising in animal models, data from early-phase clinical trials have not been encouraging. These studies have documented the ability of vaccines to prime T cells, but these responses have not correlated with therapeutic effectiveness. Because the location of T-cell activation and effector activity may impact T-cell function significantly, attention has been placed on the tumor microenvironment. There is now evidence that T-cell responses may be altered in the setting of established tumor growth. To alter the balance between local immune responses and tumor growth, a vaccine regimen using direct intratumoral delivery of poxvirus vaccines expressing costimulatory molecules was developed. An initial series of clinical trials has documented the safety profile of this approach and has suggested that local and systemic immunity can be induced. Herein, we review the general rationale for this approach, discuss the results of ongoing clinical trials, and highlight critical questions that can be addressed in well-designed preclinical and clinical studies.