We study the influence of polarization effects in streaking by combined atto-and femtosecond pulses. The polarization-induced terms alter the streaking spectrum. The normal streaking spectrum, which maps to the vector potential of the femtosecond pulse, is modified by a contribution following the field instead. We show that polarization effects may lead to an apparent temporal shift, that needs to be properly accounted for in the analysis. The effect may be isolated and studied by angle-resolved photoelectron spectroscopy from oriented polar molecules. We also show that polarization effects will lead to an apparent temporal shift of 50 as between photoelectrons from a 2p and 1s state in atomic hydrogen.PACS numbers: 42.50. Hz, 42.65.Re, 32.60.+i The attosecond (1 as = 10 −18 s) defines the natural time scale for electronic motion within atoms and molecules, just as the femtosecond (fs) is the natural time scale for nuclear motion in molecules. This is one reason for the large current interest in attosecond science [1]. Photoelectron spectroscopy with combined extreme ultraviolet attosecond (xuv) and few-cycle nearinfrared (ir) pulses is used both to characterize attosecond pulses [2,3], few-cycle laser pulses [4] and to measure ultrafast electron dynamics [5,6,7,8]. Attosecond streaking is a very promising tool for time-resolved measurements with sub-fs resolution. In attosecond streaking, one exploits that the attosecond xuv pulse is very short compared to the optical period of the assisting ir field. Then, the electrons released by the xuv pulse are all released at a definite phase of the ir pulse and obtain a momentum change due to the propagation in the ir field given classically by ∆ k(τ ) = − ∞ τ F (t)dt = − A(τ ), τ is the time of ionization by the attosecond pulse, F (t) is the electric field and A(t) the vector potential [atomic units (a.u.) with = e = a 0 = m e = 1 are used throughout unless indicated otherwise]. To obtain a clear streaking spectrum, the ir pulse should be sufficiently intense that the streaking momentum change is clearly seen, but still sufficiently weak not to excite or ionize the target. A schematic presentation of a streaking experiment is shown in Fig. 1.In this work, we show that if the target is polarized by the ir pulse, the conventional streaking spectrum, mapping to the vector potential of the ir field, is modified by a dipole term proportional to the field and a polarizability term quadratic in the field. Two different curves are plotted in the bottom of Fig. 1, corresponding to two different signals, in this case electrons released from different orbitals. The shift between the two curves may be interpreted as a temporal delay due to difference in emission time between the two signals. The linear shift following F is out of phase with the vector potential and when the phase of the streaking signal is used as a measure of emis- sion time, such a shift will manifest itself as an apparent time-delay, that needs to be properly accounted for in the analysis. We show that appa...