Cyclic deformation experiments by the Mughrabi–Ackermann technique (predeformation with plastic shear-strain amplitude ε_pl⁰⁄2=1×10⁻³ at 530 K till saturation) have been performed on a molybdenum single crystal of ultra-high purity (residual resistivity ratio RRR₀≈4×10⁵; angle χ=29° between the plane of maximum resolved shear stress and the ⟨111⟩{110} slip system with largest Schmid factor), covering the temperature range 123K≤T≤460K and seven logarithmically spaced strain rates (1.0×10⁻⁶s⁻¹≤|\dotε_pl|≤1.0×10⁻³s⁻¹). The analysis of the effective flow stress σ^* in terms of the kink pair-formation theory gave the same results as the earlier experiments on Mo crystals of the same high purity but with smaller χ [L. Hollang, M. Hommel, and A. Seeger, phys. stat. sol. (a) 160 (1996) 329] both for the energy of two isolated kinks in ⟨111⟩a₀⁄2 screw dislocations, 2H_k=1.27 eV, and for the kink height a, which agreed with the distance a_{112} between neighbouring Peierls valleys on the {112} planes. The transition between the elastic-interaction approximation and the line-tension approximation of the kink pair-formation theory, which is responsible for the strain rate-dependent “upper bend” in the σ^*-T relationship, occurred at \hatσ^*=110 MPa. In agreement with theoretical predictions, at effective stresses less than \hatσ^* the flow stress was the same in tension and compression, i.e., there was no flow-stress asymmetry. Below the upper-bend temperature, \hatT, the flow-stress asymmetry Δσ (=algebraic sum of the positive flow stress in tension and the negative flow stress in compression) increased with decreasing temperature, was positive in agreement with the “twinning–anti-twinning rule”, and reached a plateau at the lowest temperatures investigated. Since for reasons of symmetry, slip on {110} cannot give rise to a flow-stress asymmetry, together with a=a_{112} this result confirms that over the entire temperature range investigated the elementary slip steps take place on {112} planes. In the other two ultrahigh-purity crystals investigated (χ=0°, 21°), the flow-stress asymmetry has the opposite sign (in violation of the “twinning–anti-twinning rule”) and, at low temperatures, much smaller absolute values than at χ≈30°. Furthermore, Δσ<0 extends to temperatures well above \hatT. Hence there must exist, in addition to the “twinning–anti-twinning asymmetry”, a further asymmetry-producing mechanism (dubbed “straining asymmetry”), for which the following model is proposed. In the bcc metals the screw-dislocation cores of lowest energy have {110} slip planes; the configuration with {112} slip planes is populated by thermal activation. Tensile strain normal to the {112} slip plane reduces the difference in the line energy of the two configurations and, as a consequence, the Peierls potential on the {112} slip plane. Compression strain has the opposite effect.