A standard procedure is needed for obtaining smoke toxic potency data for use in fire hazard and risk analyses. Room fire testing of finished products is impractical, directing attention to the use of apparatus that can obtain the needed data quickly and at affordable cost. This report presents examination of the fourth of a series bench-scale fire tests to produce data on the yields of toxic products in both pre-flashover and post-flashover flaming fires. The apparatus is the ISO 5660-1 / ASTM E 1354 cone calorimeter, modified to have an enclosure and a gas delivery system allowing variable oxygen concentration. The test specimens was cut from finished products that were also burned in room-scale tests: a sofa made of upholstered cushions on a steel frame, particleboard bookcases with a laminated finish, and household electric cable. Initially, the standard test procedure was followed. Subsequent variation in the procedure included reducing the supplied oxygen volume fraction to 0.18, 0.16, and 0.14, reducing the incident heat flux to 25 kW/m2, and reducing the gas flow rate by half. The yields of CO2 CO, HCl, and HCN were determined. The yields of other toxicants (NO, NO2, formaldehyde, and acrolein) were below the detection limits, but volume fractions at the detection limits were shown to be of limited toxicological importance relative to the detected toxicants. In general, performing the tests at the reduced oxygen volume fraction led to small increases on the toxic gas yields. The exceptions were an increase in the CO yield for the bookcase at 0.14 oxygen volume fraction. Reducing the incident heat flux had little effect on the toxic gas yields, other than increasing variability. Reducing the gas flow rate reduced the limits of detection by half, but also resulted in reduced gas yields at lower oxygen volume fractions. In none of the procedure variations did the CO yield approach the value of 0.2 found in real-scale post flashover fire tests.