This paper details the development and verification of ROSESIM, a computer simulation model of the growth of `Royalty' roses (Rosa hybrida L.) based on experimentally observed growth responses from pinch until flowering under 15 combinations of constant photosynthetic photon flux (PPF), day temperature (DT), and night temperature (NT). Selected according to a rotatable central composite design, these treatment combinations represent commercial greenhouse conditions during the winter and spring in the midwestern United States; each selected condition was maintained in an environmental growth chamber having 12-hour photoperiods. ROSESIM incorporates regression models of four flower development characteristics (days from pinch to visible bud, first color, sepal reflex, and flowering) that are full quadratic polynomials in PPF, DT, and NT. ROSESIM also incorporates mathematical models of nine plant growth characteristics (stem length and the following fresh and dry weights: stem, leaf, flower, and total) based on data recorded every 10 days and at flowering. At each design point, a cubic regression in time (days from pinch) estimated the plant growth characteristics on intermediate days; then difference equations were developed to predict the resulting daily growth increments as third-degree polynomial functions of days from pinch, PPF, DT, and NT. ROSESIM was verified by plotting against time each simulated plant growth characteristic and the associated experimental observations for the eight factorial design points defining the region of interest. Moreover, one-way analysis of variance procedures were applied to the differences between ROSESIM predictions and the corresponding observed means for all 15 treatment combinations. At 20 days from pinch, significant differences (P < 0.05) were observed for all nine plant growth characteristics. At 30 and 40 days from pinch, only flower fresh and dry weights yielded significant differences; at flowering, none of the 13 selected responses yielded significant differences. These graphical and statistical comparisons provide good evidence of ROSESIM's ability to predict the growth response of `Royalty' roses over a wide range of constant environmental conditions.