Our previous measurements in the P and D series in 133Cs have been extended to a number of higher lying levels, populated in a two-step excitation process involving an rf lamp and a CW dye laser. The D states were studied with the level-crossing technique, while the optical-double-resonance method was used for the investigation of the P states. For the magnetic dipole interaction constant a, the Landé gJ factor, and the tensor polarizability α2 we obtain12 2P3/2: a = 1.10(3) MHz, gJ = 1.3340(15)13 2P3/2: a = 0.77(5) MHz, gJ = 1.3337(20)7 2D3/2: |a| = 7.4(2) MHz14 2D3/2: |a| = 0.425(7) MHz15 2D3/2: |a| = 0.325(8) MHz16 2D3/2: |a| = 0.255(12) MHz17 2D3/2: |a| = 0.190(12) MHz18 2D3/2: |a| = 0.160(10) MHz13 2D5/2: α2 = 19(1) GHz/(kV/cm)214 2D5/2: α2 = 37(2) GHz/(kV/cm)215 2D5/2: α2 = 70(4) GHz/(kV/cm)216 2D5/2: α2 = 120(6) GHz/(kV/cm)217 2D5/2: α2 = 199(10) GHz/(kV/cm)218 2D5/2: α2 = 323(16) GHz/(kV/cm)2Ab initio calculations of many-body perturbation type, including polarization effects to all orders, have been performed in order to obtain theoretical values of the hyperfine structure constants. The resulting a factors are listed together with the corresponding experimental values for most S, P, and D states in 133Cs measured so far. The discrepancy between the experimental and theoretical values is mainly due to the omission of the correlation effect, which is essentially as important as the polarization effect. Still, for the D states the polarization effect is large enough to yield negative values for all the 2D5/2 states, which is in accordance with the experimental results. Semiempirical values of the a factors have also been obtained using the Fermi-Segré-Goudsmit formula, and the results are compared with the experimental values. The measured values of α2 agree well with values calculated with the Coulomb approximation due to Bates and Damgaard.