Mozilla Firefox 3.0.13 and earlier, 3.5, 3.6 a1 pre, and 3.7 a1 pre; SeaMonkey 1.1.17; and Mozilla 1.7.x and earlier do not properly block data: URIs in Refresh headers in HTTP responses, which allows remote attackers to conduct cross-site scripting (XSS) attacks via vectors related to (1) injecting a Refresh header that contains JavaScript sequences in a data:text/html URI or (2) entering a data:text/html URI with JavaScript sequences when specifying the content of a Refresh header. NOTE: in some product versions, the JavaScript executes outside of the context of the HTTP site.

Argument injection vulnerability involving Mozilla, when certain URIs are registered, allows remote attackers to conduct cross-browser scripting attacks and execute arbitrary commands via shell metacharacters in an unspecified URI, which are inserted into the command line when invoking the handling process, a similar issue to CVE-2007-3670.

Cross-site scripting (XSS) vulnerability in Mozilla 1.7.12 and possibly earlier, Mozilla Firefox 1.0.7 and possibly earlier, and Netscape 8.1 and possibly earlier, allows remote attackers to inject arbitrary web script or HTML via the -moz-binding (Cascading Style Sheets) CSS property, which does not require that the style sheet have the same origin as the web page, as demonstrated by the compromise of a large number of LiveJournal accounts.

Firefox and Mozilla can associate a cookie with multiple domains when the DNS resolver has a non-root domain in its search list, which allows remote attackers to trick a user into accepting a cookie for a hostname formed via search-list expansion of the hostname entered by the user, or steal a cookie for an expanded hostname, as demonstrated by an attacker who operates an ap1.com Internet web site to steal cookies associated with an ap1.com.example.com intranet web site.

The XMLHttpRequest object in Mozilla 1.7.8 supports the HTTP TRACE method, which allows remote attackers to obtain (1) proxy authentication passwords via a request with a "Max-Forwards: 0" header or (2) arbitrary local passwords on the web server that hosts this object.

Mozilla allows remote attackers to cause a denial of service (CPU consumption) via a Javascript BODY onload event that calls the window function.

The browser user interface in Firefox before 1.0.5, Mozilla before 1.7.9, and Netscape 8.0.2 and 7.2 does not properly distinguish between user-generated events and untrusted synthetic events, which makes it easier for remote attackers to perform dangerous actions that normally could only be performed manually by the user.

Firefox before 1.0.5, Thunderbird before 1.0.5, Mozilla before 1.7.9, Netscape 8.0.2, and K-Meleon 0.9 runs XBL scripts even when Javascript has been disabled, which makes it easier for remote attackers to bypass such protection.

The InstallTrigger.install method in Firefox before 1.0.5 and Mozilla before 1.7.9 allows remote attackers to execute a callback function in the context of another domain by forcing a page navigation after the install method has been called, which causes the callback to be run in the context of the new page and results in a same origin violation.

Firefox before 1.0.5, Mozilla before 1.7.9, and Netscape 8.0.2 and 7.2 allows remote attackers to cause a denial of service (access violation and crash), and possibly execute arbitrary code, by calling InstallVersion.compareTo with an object instead of a string.

Firefox before 1.0.5 and Mozilla before 1.7.9 allows a child frame to call top.focus and other methods in a parent frame, even when the parent is in a different domain, which violates the same origin policy and allows remote attackers to steal sensitive information such as cookies and passwords from web sites whose child frames do not verify that they are in the same domain as their parents.

Firefox before 1.0.5 and Mozilla before 1.7.9 does not clearly associate a Javascript dialog box with the web page that generated it, which allows remote attackers to spoof a dialog box from a trusted site and facilitates phishing attacks, aka the "Dialog Origin Spoofing Vulnerability."

Firefox before 1.0.5, Mozilla before 1.7.9, and Netscape 8.0.2 does not properly verify the associated types of DOM node names within the context of their namespaces, which allows remote attackers to modify certain tag properties, possibly leading to execution of arbitrary script or code, as demonstrated using an XHTML document with IMG tags with custom properties ("XHTML node spoofing").

Firefox before 1.0.5 and Mozilla before 1.7.9 does not properly clone base objects, which allows remote attackers to execute arbitrary code by navigating the prototype chain to reach a privileged object.

Mozilla 1.7.8, Firefox 1.0.4, Camino 0.8.4, Netscape 8.0.2, and K-Meleon 0.9, and possibly other products that use the Gecko engine, allow remote attackers to cause a denial of service (application crash) via JavaScript that repeatedly calls an empty function.

Firefox before 1.0 and Mozilla before 1.7.5 display the SSL lock icon when an insecure page loads a binary file from a trusted site, which could facilitate phishing attacks.

Heap-based buffer overflow in MSG_UnEscapeSearchUrl in nsNNTPProtocol.cpp for Mozilla 1.7.3 and earlier allows remote attackers to cause a denial of service (application crash) via an NNTP URL (news:) with a trailing '\' (backslash) character, which prevents a string from being NULL terminated.

Firefox before 1.0 and Mozilla before 1.7.5 allows inactive (background) tabs to launch dialog boxes, which can allow remote attackers to spoof the dialog boxes from web sites in other windows and facilitate phishing attacks, aka the "Dialog Box Spoofing Vulnerability."

Firefox before 1.0 and Mozilla before 1.7.5 allow inactive (background) tabs to focus on input being entered in the active tab, as originally reported using form fields, which allows remote attackers to steal sensitive data that is intended for other sites, which could facilitate phishing attacks.

Mozilla 1.5 through 1.7 allows a CA certificate to be imported even when their DN is the same as that of the built-in CA root certificate, which allows remote attackers to cause a denial of service to SSL pages because the malicious certificate is treated as invalid.