Sustained attention (SA) is an endogenous form of attention that emerges in infancy and reflects cognitive engagement and processing. SA is critical for learning and has been measured using different methods during screen-based and interactive contexts involving social and nonsocial stimuli. How SA differs by measurement method, context, and stimuli across development in infancy is not fully understood. This 2-year longitudinal study examines attention using one measure of overall looking behavior and three measures of SA--mean look duration, percent time in heart rate-defined SA, and heart rate change during SA--in N = 53 infants from 1 to 24 months across four unique task conditions: social videos, nonsocial videos, social interactions (face-to-face play), and nonsocial interactions (toy engagement). Results suggest that developmental changes in attention differ by measurement method, task context (screen or interaction), and task stimulus (social or nonsocial). During social interactions, overall looking and look durations declined after age 3-4 months, whereas heart rate-defined attention measures remained stable. All SA measures were greater for videos than for live interaction conditions throughout the first 6 months, but SA to social and nonsocial stimuli within each task context were equivalent. In the second year of life, SA measured with look durations was greater for social videos compared to other conditions, heart rate-defined SA was greater for social videos compared to nonsocial interactions, and heart rate change during SA was similar across conditions. Together, these results suggest that different measures of attention to social and nonsocial stimuli may reflect unique developmental processes and are important to compare and consider together, particularly when using infant attention as a marker of typical or atypical development.