Typically, balance training has been used as an intervention paradigm either as static or as reactive balance training. Possible differences in functional outcomes between the two modalities have not been profoundly studied. The objective of the study was to investigate the specificity of neuromuscular adaptations in response to two balance intervention modalities within test and intervention paradigms containing characteristics of both profiles: classical sensorimotor training (SMT) referring to a static ledger pivoting around the ankle joint vs. reactive balance training (RBT) using externally applied perturbations to deteriorate body equilibrium. Thirty-eight subjects were assigned to either SMT or RBT. Before and after four weeks of intervention training, postural sway and electromyographic activities of shank and thigh muscles were recorded and co-contraction indices (CCI) were calculated. We argue that specificity of training interventions could be transferred into corresponding test settings containing properties of SMT and RBT, respectively. The results revealed that i) postural sway was reduced in both intervention groups in all test paradigms; magnitude of changes and effect sizes differed dependent on the paradigm: when training and paradigm coincided most, effects were augmented (P<0.05). ii) These specificities were accompanied by segmental modulations in the amount of CCI, with a greater reduction within the CCI of thigh muscles after RBT compared to the shank muscles after SMT (P<0.05). The results clearly indicate the relationship between test and intervention specificity in balance performance. Hence, specific training modalities of postural control cause multi-segmental and context-specific adaptations, depending upon the characteristics of the trained postural strategy. In relation to fall prevention, perturbation training could serve as an extension to SMT to include the proximal segment, and thus the control of structures near to the body's centre of mass, into training.