The first Avrocar, #58-7055, rolled out of the Avro Malton factory in May 1959. From 9 June to 7 October 1959, it was tested in a static hover rig. Unfortunately, hot gas from the exhaust was found to mix back into the intakes in hover, reducing engine thrust. In addition, the fan generated lift only from a small area of its surface, lowering overall thrust available. The ducts also proved to have higher losses than expected, and a series of modifications were not able to solve this problem to any large degree. These problems reduced maximum lift at higher altitudes out of the ground effect to 3,150 lb (1,430 kg), less than the empty weight of the aircraft at 4,285 lb (1,944 kg). This meant the aircraft would be incapable of hovering out of the ground effect. Following these tests, the vehicle was sent to NASA Ames for a wind tunnel testing.
The second, #59-4975, was completed August 1959. On 29 September, the first attempt to hover was made with the Avrocar tethered to the ground. After the vehicle became airborne, an uncontrollable roll and pitch-coupled oscillation started that forced each of the three wheels into the ground in turn. The pilot, W.D. "Spud" Potocki, immediately shut down all engines. Changes were made to the stability system to provide more control authority, while new tethers were investigated to improve the ability to control these sorts of problems. As testing continued it became clear that the problem was inherent to the design, and the engineers started referring to the effect as "hubcapping," so-named as it appeared similar to a hubcap spun on the ground.
Investigations into the effect revealed what was causing the problem. While in the ground effect, the high-pressure air under the craft was trapped, filling the entire area and thus providing a stable base. When the craft rose out of the ground effect, the air formed itself into a single narrow column, described by Frost as "tree trunking". At intermediate altitudes the craft would momentarily transition from one regime to the other, during which time one side of the vehicle would be entirely supported while the support was disappearing under the other. This led to a strong pitching motion towards the unsupported side. As soon as this occurred that side would approach the ground and re-establish the supporting air, while the other side would then be raised above this limit. This process would repeat itself, with the craft rolling from side to side. Modifications were carried out in order to try to solve the problem. Eventually a series of 52 holes were drilled in the bottom of the vehicle, located radially three feet from the center. These were to provide a central jet to stabilize the ground cushion.
With these modifications complete and apparently working, the first completely free flight occurred on 12 November 1959. This test proved the nozzle control system unacceptable. The spoilers were intended to direct the air out over the top or bottom of the annular flap, out the bottom only during hover, but over the top and bottom during forward flight. The idea was that when the flap was positioned in order to provide control, the lift would be lowered on one side and raised on the other. Lift was indeed lowered on one side, but sadly the lift did not improve on the other, so every control input resulted in a loss of altitude. After five flights, testing was temporarily halted on 5 December 1959, by which time the Avrocar had logged 18.5 hours of test time in total.
A completely new design was installed over the winter. The original spoilers were removed and replaced by a single ring below the annular flap. The ring shifted in relation to the craft under control input, "sealing off" the gap on one side while opening it on the other. Tests continued in January 1960, and appeared to work much better. However, while the new control system improved the hovering qualities, the craft now became unstable at higher speeds above 30 knots (56 km/h). The first Avrocar at Ames was similarly modified, and, in April 1960, it was tested in their 40 x 80 ft (24 m) wind tunnel. The problem became clear; the ring blocked so much of the overall engine thrust that the overall power was greatly reduced. As the craft sped up, the airflow on the underside reduced the recirculation, reducing the lift due to airflow over the upper surface. This was somewhat unexpected; recirculation had been considered a "bad thing" due to the loss of engine thrust, and its positive effects on lift do not appear to have been appreciated. In the end, in order to maintain lift at higher forward speeds the craft had to be pitched to the end of its control capability.
Avro was convinced that the concept was still workable, and proposed a new program for major rework of the propulsion and control system. Instead of the single annular triangular flap and spoilers, or the later ring control, the new system included two separate control systems for hover and forward flight, combined into a single nozzle. For hover, a series of "transition doors" were opened into the nozzles, blocking them off and re-directing the flow downward under the aircraft. Control during this regime was provided by moving the outer portion of the flap to "focus" the flow. At higher speeds, the doors were closed, allowing the air to flow out from the edge of the aircraft, where a series of simple flap-like controls were located. The new control system covered the rear 3/4's of the aircraft's outer circumference; the front section featured the hovering controls only.
Modifications were completed on the Ames model and testing resumed in April 1961. The new design demonstrated much better control in hover and considerably improved lift. The vehicle was now able to travel at up to 100 knots (190 km/h), a great improvement over the 30 knots (56 km/h) previously reached. However, it remained unstable in pitch, and exhibited a strong nose-up trim. NASA engineers attempted to modify this with a T-tail, but this proved to sit within the airflow of the turborotor and did not help. Frost's team considered two new designs, one with a large vertical tail and one with a wing with tip mounted verticals —"winglets." Both designs used two 2,700 lbf (12 kN) thrust General Electric J85 turbojets and increased the turborotor diameter from five to six ft.
On 9 June 1961, a second USAF/NASA flight evaluation of the Avrocar was conducted on the similarly modified second prototype at the Avro facility. During these tests, the vehicle reached a maximum speed of 20 knots (37 km/h) and showed the ability to traverse a ditch six feet across and 18 inches (460 mm) deep. Flight above the critical altitude proved dangerous if not nearly impossible due to inherent instability. The flight test report further identified a range of control problems.