This paper argues that growth theory needs a more general notion of "regularity" than that of exponential growth. We suggest that paths along which the rate of decline of the growth rate is proportional to the growth rate itself deserve attention. This opens up for considering a richer set of parameter combinations than in standard growth models. And it avoids the usual oversimplistic dichotomy of either exponential growth or stagnation. Allowing zero population growth in three different growth models (the Jones R&D-based model, a learning-by-doing model, and an embodied technical change model) serve as illustrations that a continuum of "regular" growth processes fill the whole range between exponential growth and complete stagnation.