We consider the internal dynamics of the polymer molecule which is injected in the chaotic flow with strong mean shear component. The flow geometry corresponds to the recent experiments on the elastic turbulence (Groisman, Steinberg 2000). The passive polymer in such flows experiences aperiodic tumbling. We present a detailed study of the statistical properties of such polymer dynamics. First we obtain the stationary probability distribution function of the polymer orientation. Secondly we find the distribution of the time periods between consequent events of tumbling, and finally we find the tails of the polymer size distribution.PACS numbers: 83.80. Rs,47.27.Nz Hydrodynamics and rheology of dilute polymer solutions has attracted much theoretical and experimental attention recently. Addition of small amount of polymers to ordinary liquid leads to crucial changes of liquid properties. One of the most famous effects of this type is the phenomenon of drag reduction. The addition of few parts per million of long-chain polymer molecules produces a dramatic reduction of the friction drag. Although this effect was first observed by Toms in 1949 [1], there is still no rigorous theory, explaining the phenomena. The qualitative description was proposed by Lumley [2,3], but no quantitative theory is available. Another spectacular phenomena, observed in dilute polymer solutions is the effect of elastic turbulence, discovered recently by Groisman and Steinberg [4,5]. In this experiment a chaotic fluid motion was observed in the system with small Reynolds number Re ≪ 1. Obviously, such behavior can not be observed in Newtonian liquids, where the flow should be laminar, so the chaotic flow is generated by the elastic instabilities of polymer solution. The dynamics of polymers and possible mechanisms, explaining the chaotic state were studied in the recent theoretical works [6,7,8]. It was proposed that elastic instabilities occur because of elongation of a single polymer in external flows. The analysis of the single polymer dynamics in chaotic flows, shows that the transition between two qualitatively different types of behavior of polymers can be observed in such system. This transition is called coil-stretch transition. It separates the dynamics in weak flows, where polymer molecules remain in coiled state most of the time, and strong flows, where the molecules become substantially elongated. The measure of the flow strength is given by the Weissenberg number which is the product of the characteristic velocity gradient and the polymer relaxation time. More precisely the Weissenberg number is defined as Wi = λτ , where λ is the largest Lyapunov exponent, associated with the flow, and τ is the relaxation time of the slowest polymer excitation mode. The coil-stretch transition occurs at Wi = 1. With the development of novel optical methods a number of high quality experimental observations focusing on resolving dynamics of individual polymers (DNA molecules) subjected to a non-homogeneous flow have been reported [9,10,11,1...