Tractor overturns are a major cause of agricultural-worker deaths each year. These deaths and serious injuries may have been prevented if the tractors had been equipped with rollover protective structure (ROPS), and the operator was wearing a seat belt. Many tractors manufactured prior to 1970 did not have ROPS as an option, and thus the axle mounts were not designed to structurally support a ROPS during an overturn. If ROPS were available for these pre-ROPS tractors, then rollover protection will be more readily available and lives could be saved. A database of tractors in use on U.S. farms (Myers and Snyder, 1995) was used to define the major axle categories for pre-ROPS tractors. The first major axle category includes the axle design for the John Deere A, B, G, 50, 60, 70, 520, 620, 720, 530, 630, and 730 tractors. This axle category consists of about 920,000 tractors sold with approximately 150,000 still in operation. The axle housings, and thus ROPS mounting locations are very similar for these tractor models. Axle drawing for these tractors have been obtained from Deere and Co. Axle housing dimensions and bolt hole locations were identified to assist in the ROPS construction guidelines. The second major axle category selected covers the Farmall H, M, Super H, Super K 300, 400, 350, 450, 460 tractors. Again, these model tractors have very similar axle housings and axle housing drawings were constructed. Although ROPS for some of these tractors are listed as being available through Saf-T-Cab, Saf-T-Cab would like a more economical ROPS design (two post). Also, this tractor axle category makes up a large number of tractors in operation (approximately 278,000). Again, axle housing dimensions and bolt hole locations were identified to assist in the ROPS construction guidelines. Two-post ROPS have been designed, constructed and tested for the John Deere A and Farmall M tractors. Energy calculations and allowable deflections were used along with the exposure criteria model to design the ROPS dimensions and select the material. The exposure criteria model was modified to assist in the ROPS design for the two tractors. The static lateral, longitudinal and vertical tests were conducted using the ROPS testing apparatus to ensure the requirements stated in ASAE S519 were satisfied. Both John Deere A and Farmall M tractors were purchased and modified for field upset testing. The modification includes structural support, battery protection, gas tank redesign and installation of a pneumatic power source for starting, braking and clutching. Pneumatic tanks, control valves and cylinders were installed and are radio controlled to provide remote operation of the tractor. The ROPS deflection measuring system was installed. The lateral and longitudinal field upset sites were constructed in accordance with ASAE S519. Field upset testing was conducted for the John Deere A and the Farmall M in 1996 and 1997, respectively. The designed ROPS for both tractors satisfied the ASAE S519 test requirements. Initially, vibrational loading was a concern when mounting ROPS on pre-ROPS tractors. However, after discussions with Dan Lockie of Saf-T-Cab, Dale Baker and Joe Oliver of Case IH, and Murray Madsen of Deere and Company, axle strength during the rear upset tests is more of a concern. The concern in the industry is the ultimate strength of the tractor axle housing when subjected to rear (longitudinal) loading. In rear loading, if the axle housing fails, the housing will rotate with respect to the axle and the operator will be crushed. To address this concern, an axle strength test apparatus was constructed. This apparatus consists of a 4-inch diameter, 24-inch stroke hydraulic cylinder attached to a 7- foot length I-beam. A torsional load is placed on the axle housing until failure or the 20,000 pound limit of the load cell is reached. The longitudinal torsional strength of 17 pre-ROPS axle housings (8 pre-ROPS model tractors) were tested. The longitudinal strength test were repeated four times for the Ford 8N and Farmall M pre-ROPS axle housings. Torsional axle housing strength test results of John Deere A, Ford 2N, Farmall H, C, 450 and 460 pre-ROPS axle housings were also obtained. Guidelines of ROPS design which focus on axle housing support need to consider the design margin between longitudinal yield torque of the axle housing and maximum torque subjected during ASAE Standard S519 longitudinal static test. This design margin is determined to evaluate the appropriateness of the axle housing to support the designed ROPS. For example, the design margins for the Farmall M ROPS/axle housing combination is described. Static longitudinal tests conducted in accordance with ASAE S519 at Colorado State University revealed an estimated maximum torque on the axle housing as 36,000 Newton-meters. The strength torques of the four Farmall Maxie housings were 51,436, 54,857,46,520 and 49,574 Newton-meters. Their design margin are 1.43, 1.52, 1.29 and 1.38. These design margin calculations are only valid for the ROPS design tested. A change in ROPS design would likely produce a change in the design margin. The design margin of the Farmall M axle housing has a population mean of 1.41 with the standard deviation of 0.10. The 95% confidence interval of the design margin is determined to range from 1.25 to 1.57. The probability of the design margin less than one (the strength torque of the axle housing is equal to the maximum torque applied during ASAE Standard S519 longitudinal static test) is less than 0.35%. The test results indicate that the Farmall M axle housings tested can successfully support the ROPS design tested. Initial satisfactory results were found for the Ford 8N and John Deere A axle housings, encouraging commercial ROPS design for these tractors. These guidelines can be used to evaluate the appropriateness of axle housing support during ROPS design for pre-ROPS tractors. A Fortran program was developed to determine if the operator clearance zone is exposed to the ground surface as the ROPS is deflected during the static test. The model resultswere compared to actual RapS static tests to evaluate its accuracy. In actual testing the clearance zone was exposed to the ground plane at a horizontal RapS load point deflection within 24 mm of the model prediction. This model provides guidelines in RapS design (particularly sizing and dimensioning) when considering the evaluation of the exposure criteria as defined in ASAE S519.