SCIENTIFIC ABSTRACT Most objects with cosmological redshift (z > 0.5) have been found by means of their radio emission (radio loud quasars and radio galaxies) or due to their optical emission lines (radio quiet quasars). Very little is known about the FIR spectra of these objects since IRAS could obtain spectra only for the 50 brightest members of this class. On the other hand, it is just the range between 20 and 200 microns where one expects the thermal emission to overrun the non-thermal radiation dominating the radio regime. Thus FIR spectra should show both the shortest wavelengths of non-thermal and the longest wavelengths of thermal continuum emission which are of great importance for a theoretical understanding of the various emission pro- cesses and their origin in the cores of AGNs. We propose to measure the spectrum of several carefully selected quasar samples between 5 and 200 microns with ISOPHOT-P and ISOPHOT-C in order to identify the different components. This will be done by quasi-simultaneous multi-band photometry. Polarization and variability studies which can even clearer distinguish between synchrotron and other components will be performed for a few key objects. Evolutionary trends will be investigated by measuring quasars in the redshift range between 0.1 and 4.5. Observations of a sample of radio gal- axies from the 3C-Catalogue will test whether the unified schemes are viable which interpret radio galaxies as quasars seen under distinct angles. The total spacecraft time of the proposal is 35 hours with a mean observing efficiency of 45 %. OBSERVATION SUMMARY The continuum emission of quasars between 4.8 and 200 microns will be traced with ISOPHOT-P in six filters (P_4.85, P_7.3, P_12.8, P_20, P_60, P_100) and ISOPHOT-C (C_160) in order to obtain an almost complete and simultaneous energy distribution. The apertures are 23arcsec (4.85 to 20 microns) and 99arcsec (60 and 100 microns), the chopper throw is 120arcsec for the P-filters and 180 arcsec for the C-filter. A total observing time of 855sec per low redshift quasar will result in a S/N of about 3 at 4.8 micron and more than 100 at 160 microns (on source integration times: 32s in P_4.85, 64s in P_7.3, 16s in the remaining bands). The high redshift objects are much fainter and will be observed only in P_12.8 (256s on-source), P_20 (256s), P_60 (128s), P_100 (32s) and C_160(64s) resulting in a total time of 1957 sec per object. The observations comprise various QSO samples which are selected by the visibility constraints of ISO, i.e. 3CR sources (30 objects, 6.4h), 3Jy catalog + BLL (35 objects, 7.6h), high-z QSOs (9 objects, 4.9h) and PG sources (24 objects 5.2h). The multi-band photometry should be repeated for 3 radio-loud QSOs (3C273, FR_0133+476, FR_0208-512 (spring) or 3C273, 3C345, BL_2155-152 (autumn)) within weeks and for 2 optically selected QSOs (1613+658, 1634+706 (autumn launch) or 0946+301, 0953+414 (spring launch)) within intervals of several months. In total 35 measurements (8.3h) are proposed to study the variability on different time scales. Four sources will be examined by polarimetric observations at 90 microns in order to distinguish between thermal and non-thermal contributions. In order to achieve a polarization error of less than 0.5%, on average 850sec of observing time per source are required.