A new electrooptical apparatus has been used to characterize the dichroism decay time constants for a collection of nine blunt-ended DNA restriction fragments in the range of chain lengths from 41 to 256 base-pairs at physiological salt concentrations. The experimental data show an increase of rotational diffusion coefficients, when the monovalent salt concentration is increased from a few mM, used previously for standard electrooptical experiments, to the range of salt concentrations around 100 mM. The presence or absence of 10 mM Mg2+ in a buffer with 100 mM NaCl does not induce any large change of the rotational diffusion. Bending of double helices is reflected by a fast component in the dichroism decay for fragments greater than or equal to 90 bp; the time constant of the first bending mode is 7-9% relative to the time constant of overall rotational diffusion for fragments with 90 to 179 bp at the temperatures 2, 10 and 20 degrees C. Interpretation of the overall rotational diffusion time constants by different models on the hydrodynamics of flexible polymer chains leads to diverging values of the persistence length. The most accurate description is expected from a combination of the rotational diffusion coefficient for rigid rods given by Tirado and Garcia de la Torre (J. Chem. Phys. 73 (1980) 1986) with correction factors derived from Monte Carlo simulations (P.J. Hagerman and B.H. Zimm, Biopolymers 20 (1981) 1481). This model leads to 'average' values of the persistence length of 440, 400 and 380 A at the temperatures 20, 10 and 2 degrees C, respectively (in 110 mM Na+ and 10 mM Mg2+, pH 7.0); the hydrodynamic radius of the helix is approx. 12.5 A. The persistence lengths measured at various monovalent salt concentrations can be represented as a linear function of the reciprocal square root of the ionic strength. The rotational time constants measured for individual fragments at physiological salt show clearly larger deviations from the model average than corresponding time constants measured previously at low salt; 'apparent' persistence lengths of individual fragments as well as their temperature dependence show strong variations. Thus, it is hardly possible to define a 'standard' persistence length for mixed sequences--even though the sequences used in the present investigation do not show clear deviations from standard gel mobilities. These data indicate that formation of individual, sequence-directed structures of DNA fragments is favoured under physiological salt conditions.