Jpred 3 is a secondary structure prediction server that uses the Jnet algorithm to predict the secondary structure of proteins with high accuracy. It provides predictions for over 1000 sequences per week for users in more than 50 countries. The Jnet algorithm, now updated to version 2.0, achieves an accuracy of 81.5% in blind tests for secondary structure prediction. Jpred can accept either a single protein sequence or a multiple sequence alignment (MSA) and generates predictions of secondary structure and solvent accessibility. These predictions are available in various formats, including HTML, plain text, PostScript, PDF, and via the Jalview alignment editor. The new Jpred 3 server includes significant usability improvements, such as clearer feedback on submission progress or failure, batch submission of sequences, and email summaries of results. It also updates search databases to ensure high-accuracy predictions. The server now uses a new software pipeline to stay in sync with updates to SCOP and UniProt. Jpred 3 also allows users to submit sequences via email, and results can be viewed in various formats. The server stores results for 2 days but allows users to download data for local storage. Jpred 3 can handle orphan proteins with no hits to UniRef90 by using only an HMM profile. The server also includes a new semi-automatic pipeline for creating Jnet training data, which is linked to SCOP releases for retraining. Jpred 3 improves user interaction by providing more feedback on submission progress and includes a progress meter during job execution. It also allows batch submissions and email-based results. The server is updated regularly with new sequence information to maintain high accuracy in predictions.Jpred 3 is a secondary structure prediction server that uses the Jnet algorithm to predict the secondary structure of proteins with high accuracy. It provides predictions for over 1000 sequences per week for users in more than 50 countries. The Jnet algorithm, now updated to version 2.0, achieves an accuracy of 81.5% in blind tests for secondary structure prediction. Jpred can accept either a single protein sequence or a multiple sequence alignment (MSA) and generates predictions of secondary structure and solvent accessibility. These predictions are available in various formats, including HTML, plain text, PostScript, PDF, and via the Jalview alignment editor. The new Jpred 3 server includes significant usability improvements, such as clearer feedback on submission progress or failure, batch submission of sequences, and email summaries of results. It also updates search databases to ensure high-accuracy predictions. The server now uses a new software pipeline to stay in sync with updates to SCOP and UniProt. Jpred 3 also allows users to submit sequences via email, and results can be viewed in various formats. The server stores results for 2 days but allows users to download data for local storage. Jpred 3 can handle orphan proteins with no hits to UniRef90 by using only an HMM profile. The server also includes a new semi-automatic pipeline for creating Jnet training data, which is linked to SCOP releases for retraining. Jpred 3 improves user interaction by providing more feedback on submission progress and includes a progress meter during job execution. It also allows batch submissions and email-based results. The server is updated regularly with new sequence information to maintain high accuracy in predictions.