Abstract: Enhanced geothermal systems (EGS) offer a practical route for extracting heat from hot dry rock, yet their operation can be accompanied by seismicity or microseismicity during hydraulic stimulation, after shut-in, and throughout later circulation. This seismic response is tied to the structure of hot dry rock reservoirs. These reservoirs are usually crystalline and tight, with very low matrix permeability, so fluids tend to move through hydraulic fractures, natural fractures, and fault damage zones rather than through a homogeneous porous medium. As a result, no single mechanism, whether pore-pressure diffusion, thermoelastic stressing, or poroelastic stressing, can explain the full range of EGS-induced seismic responses. Drawing on 79 publicly reported cases, together with theoretical analyses, field observations, and numerical simulations, this review examines how seismic responses and controlling mechanisms change from hydraulic stimulation to post-shut-in and circulation stages. Hydraulic stimulation accounts for the largest share of the reported cases, about 72%, whereas circulation accounts for about 26% and post-shut-in events for about 2%. These proportions, however, should not be read as a direct measure of risk. Larger events are more closely associated with the overlap between pressure perturbations, near-critical faults, preferential flow paths, and stress transfer. During hydraulic stimulation, microseismicity is generally concentrated near the wellbore or within the stimulated reservoir volume. It is mainly linked to hydraulic fracture propagation, stress concentration at fracture tips, shear slip on small natural fractures, and local stress redistribution. The situation changes when hydraulic fractures connect with faults, because pressure transmission and Coulomb stress changes may then promote fault slip. After shut-in, the end of injection does not necessarily mean that the system has stabilized. Residual pressure can continue to migrate through fracture networks toward the far field, while poroelastic rebound, fracture closure, aseismic creep, and static stress transfer from earlier events may still alter the stability of distant faults. During circulation, induced events are often long-lasting and low in magnitude. Early responses are controlled mainly by injection-production pressure differences and reactivation of existing fractures; with the advance of the cooling front, thermal contraction, fracture aperture changes, water-rock reactions, and evolving fault friction become increasingly relevant. EGS-induced seismicity is therefore better treated as a stage-dependent process. Hydraulic stimulation is dominated by fracture growth and possible fault connection; the post-shut-in period is governed more by pressure-stress redistribution; and long-term operation gradually brings thermo-hydro-mechanical-chemical coupling into play. Risk control needs to follow this evolution. Before stimulation, fault identification, in-situ stress evaluation, and well-site optimization are needed. During stimulation, staged injection, microseismic monitoring, and downhole pressure feedback can help constrain fracture growth. After shut-in, slow shut-in, stepwise pressure reduction, pressure-holding observation, or moderate flowback may be selected according to site conditions, with monitoring continued until pressure redistribution becomes relatively stable. During circulation, injection-production balance should be maintained, and microseismic, DAS, InSAR, downhole pressure, temperature, flow-rate, and fluid-chemistry data should be integrated into risk-model updating. Future models need to represent fracture and fault networks, thermo-poroelastic responses, fault frictional slip, and chemical weakening in more detail, so that traffic-light systems can move beyond empirical thresholds toward dynamic forecasts constrained by physical mechanisms.