Propagation of microorganisms is based on three fundamental processes: cell growth, DNA replication, and cell division. Although important for antibacterial drug development, these processes are poorly understood in Actinobacteria, a medically important phylum that includes Mycobacterium tuberculosis. Using microfluidic cultures and time-lapse microscopy, we studied single-cell growth, DNA replication, and cell division in the model organism Mycobacterium smegmatis. M. smegmatis is rod-shaped and grows by tip elongation in a biphasic manner due to a "new end take-off" (NETO) event. We find that pole elongation speed is increased and NETO occurs earlier in fast-growing cells than in slow-growing cells. As a consequence of variable timing of NETO, single-cell growth can be monophasic, biphasic, or triphasic. We propose that cells optimize pole growth speed and the timing of NETO to maximize their overall growth. We show that fast-growing cells initiate DNA replication earlier than slow-growing cells. We also find that single-cell growth speed is linked to cell-cycle progression, which is similar when comparing cells growing at the same speed under different conditions. We also report that multifork replication occurs when the time between DNA replication initiation events is shorter than the C period, which may occur even in slow-growing cells with interdivision times longer than the C period. Division site selection in many bacteria is governed by the nucleoid occlusion (Noc) and minicell (Min) systems. Mycobacteria do not encode homologs of these proteins and have no known mechanism for division site selection. We found that the DNA replisome and cell division ring colocalize and move together in a biphasic trajectory determined by the chromosome partitioning (Par) system. We propose a model in which Par-dependent movement of the replisome and division ring ultimately determines the site of cell division. Coordination of cell growth and division is required for cell size homeostasis. Three models of division control have been proposed: sizer (cells divide after reaching a critical size), timer (cells divide at a certain time after birth), and adder (cells add a specific amount of mass before dividing). We show that the single-cell growth model (exponential, linear, or bilinear) constrains the possible division control models. Thus, it is crucial to know the true model of single-cell growth in order to distinguish the true mechanism of cell size homeostasis.