Modeled vegetation whose morphological parameters such as leaf area index (LAI) interact with atmospheric or hydrologic variables are described as "interactive vegetation", or IV. Coupling IV into a soil-vegetation-atmosphere transfer (SVAT) scheme is becoming an active research topic. This paper discusses a strategy to couple the IV model of Dickinson et al. (1998) into the Community Land Model (CLM) and evaluate the simulations of crop growth with a two-year dataset from an agricultural site in Champaign, Illinois. The results show that CLM with IV can produce accurate simulation of observed LAI, latent and sensible heat fluxes, CO2 fluxes, soil moisture, and soil temperature for both soybean and corn. Sensitivity tests show that Vmax, the maximum leaf catalytic capacity at the canopy top that relates stomatal conductance to canopy assimilation, is important in determining LAI, CO2 flux, and latent and sensible heat fluxes. Different profiles of vertical roots distribution in the deep soil layer play an important role in the simulations of LAI and sensible and latent heat fluxes during the dry-down period. Properly specifying the planting and harvesting dates critically influences the simulations of LAI and fluxes during the green-up and green-down periods. The original dynamic leaf model of Dickinson's does not have a stem mass balance equation, thus leading to excessive allocation of carbon to leaves in the corn case. Allowing allocation to stem improves the overall simulations.