TCP-Friendly Rate Control (TFRC) was originally designed for multimedia streaming applications where continuous data was available at the sender. However, TFRC is not well-suited to the variable rate traffic presented by many modern adaptive media codecs. One way to counter this deficiency would be for the sender to continue to transmit at the media rate during periods of silence, known as padding. This use of padding can ensure acceptable application performance. However, it also degrades network performance, and decreases the usefulness of TFRC congestion control. Recent standardisation has resulted in a new revised TFRC specification. This paper describes candidate methods that were evaluated as a part of this revision and presents the first analysis of the new TFRC specification including a comparison this with the proposed Faster Restart method. It evaluates behaviour both in terms of the application performance benefit and the implications on other network traffic that share an Internet bottleneck and shows that the new methods improve the performance of bursty media. Although Faster Restart allowed TFRC to better support bursty applications, the additional gain was determined to be small when combined with the revised TFRC specification. Finally, revised TFRC is shown to remove the former incentive for padding, substantially improving the performance of other network traffic sharing a congested network.